Paper 4 370

ISBN:9780470396070

Watch video

· Video:

Buddha

From its early beginnings as DARPANET, linking a select few military and Research and Development (R&D) facilities, the Internet has grown to comprise thousands of networks and millions of users in virtually every country in the world. The Internet is the kudzuof networks, growing like the uncontrollable weed of Southern infamy.

The Internet truly was the information superhighway, long before the term was coined. While actually more of a private road, it was characterized by a sense of truly unparalleled freedom. That early network stood apart by virtue of its providing free and open access to information—that characteristic remains today as the hallmark of the Internet. The Internet is a network of networks and has become a Global Village, much as envisioned by Marshall McLuhan (1911–1980), education, philosopher, and scholar. In some sense, at least, the Internet is the information superhighway.

13.1 THE INTERNET DEFINED

In late 1983, ARPANET split into two unclassified networks: DARPANET and MILNET (MILitary NETwork). Although ARPANET officially was retired in June 1990, the Internet has survived. TCP/IP has been enhanced recently and currently enjoys wide popularity in education, research and development, and commercial applications.

The National Science Foundation founded NSFNET (National Science Foundation NETwork) in 1985 to link its supercomputer centers and to provide access to the Internet. NSFNET was a high-speed backbone network consisting of point-to-point links in a mesh configuration. The network was deployed fully in 1988, initially at 56 kbps. In 1986, NSFNET partially funded a number of regional networks, tying into the backbone. In 1992, the backbone was upgraded to T3, operating at 45 Mbps. More recently, major portions of NSFNET were upgraded to OC-3 (155 Mbps) and beyond. NSFNET officially was retired in 1995 and was replaced by the MERIT network. MERIT originally was a statewide network operated by the University of Michigan and a regional component of both NSFNET and the Internet.

13.2 INTERNET PHYSICAL TOPOLOGY

The physical topology of the Internet, ultimately, is not that important as long as connectivity is achieved. That topology also changes from day to day and varies considerably from place to place. However, and for those of you who, like me, absolutely must know at some level, the Internet physical topology consists of leased lines that connect various major computing centers through switches and routers. This leased-line network is a partially meshed network for purposes of redundancy and network resiliency. Initially, ARPANET consisted of approximately 50 custom-built BBN Packet-Switching Nodes (PSNs) scattered around the United States and Western Europe. NSFNET (1985) consisted of 56-kbps leased-line circuits that connected the NSF supercomputer centers in San Diego, California; Boulder, Colorado; Champaign, Illinois; Pittsburgh, Pennsylvania; Ithaca, New York; and Princeton, New Jersey. During this time, a great deal of Internet traffic over the NSFNET backbone became commercial in nature, prompting a number of private commercial Internet backbone providers to establish a point for the exchange of this traffic. This point was established at Wiltel offices in Santa Clara, California, and became known as the Commercial Internet eXchange (CIX).

In addition to the basic Internet backbone structure spawned by NSFNET, the major incumbent interexchange carriers—AT&T, MCI, and Sprint—built national backbones. These carriers also have either made a number of acquisitions to enhance those backbones or, more typically, have acquired backbones that they then enhance. A clear example is the acquisition of Metropolitan Fiber Systems (MFS) and its UUNET subsidiary by MCI, which then became MCI Worldcom. MCI also acquired Advanced Networks and Services (ANS), a consortium to which MCI was a party. Ultimately, MCI was acquired by Verizon. Similarly, AT&T acquired Teleport Communications Group (TCG), which previously acquired the CERFnet backbone, and AT&T ultimately was merged into SBC (nee Southwestern Bell), which retained the name AT&T. Competitive interexchange carriers, such as Level 3 Communications and 360 Networks, also offer backbone Internet transport. The incumbent LECs built major regional backbones some years ago and a number of less well-known regional carriers transport Internet traffic as well.

So far, our discussion has focused on the U.S. topology, which is appropriate, as the Internet is a U.S. invention. For all practical purposes, it exclusively was a U.S. network until the 1990s. As the Internet gradually extended outside U.S. borders, all traffic initially was routed through the U.S. infrastructure. That meant, for example, that local traffic from one user in Hong Kong to another in Hong Kong was routed through MAE West in San Jose and back to Hong Kong. In order to confine regional traffic and thereby correct this obvious inefficiency, dozens of NAPs now exist in a number of international locations. Those NAPs include The Hong Kong Internet Exchange (HKIX) in Hong Kong, the Japan Internet Exchange Co. (JPIX) in Tokyo, the Amsterdam Internet Exchange (AMS-IX) in The Netherlands, the Service for French Internet Exchange (SFINX) in Paris, the London Internet Exchange (LINX) in The United Kingdom, and the Johannesburg Internet Exchange (JINX) and Cape Town Internet Exchange (CINX) in South Africa.

To put this physical topology in some sort of reasonable order and context, consider the concept of levels. While these levels are not defined absolutely, they do serve to put the topology of the Internet in perspective. At level 1 are the interconnect points and at level 5 are the end users. In a long-haul, coast-to-coast communication in the United States, you might well work up through all five levels and back down through all five:

· Level 2 is the national backbone level. At this level are the national service providers such as AT&T, MCI, Cable & Wireless, Level 3 Communications, and Global Crossing. Level 2 providers are facilities based, as they own long-haul, high-speed transmission facilities that serve to interconnect their various high speed switches and routers. (Note: Level 3 Communications has built a very substantial IP-based network, primarily in the United States. As IP operates at Layer 3 of the OSI Reference Model, the company originally considered naming itself Layer 3 Communications but felt that name to be less attractive than Level 3, or so I was told by a number of Level 3 employees who claimed to know the underlying logic. I fully understand that this is purely anecdotal, but it is the best I can do.)

· Level 3 comprises the regional carriers. Such carriers typically operate a backbone in a single state or perhaps several states. Level 3 service providers also are facilities based, but at a lesser level than the level 2 service providers.

· Level 4 comprises ISPs, which can be small mom-and-pop companies or regional or even national in nature. ISPs may also be Level 3 or even Level 2 providers. They also may function strictly as ISPs and not own or operate any backbone transmission facilities. Examples of pure ISPs include America Online, Earth-Link, CompuServe, Prodigy, and Comcast.

· Level 5 consists of the end users. Level 5 connections to the Internet may be through a Level 4, Level 3, or even Level 2 provider. Access techniques run the full range of network options discussed in previous chapters and discussed again later in this chapter. Such techniques include leased lines, dial-up, Digital Subscriber Line (xDSL), cable modem, and Passive Optical Network (PON) [13].

13.3 INTERNET ACCESS

While the number of Internet users continues to grow at incredible rates, not everyone has access. In this Information Age, there is great concern about creating a new class system of the information haves and have-nots. In an attempt to address this issue of the digital divide, a great debate has arisen about subsidizing some level of universal service, much as was assured in the 1930s for the Public Switched Telephone Network (PSTN). With the goal of providing access to every American school by the year 2000, $ 36 million was raised from private sources in 1995—with matching funds from U.S. government sources and donations. Additionally, various carriers offered special access rates and some manufacturers donated equipment or offered special discounts [14].

According to ISP-Planet, top U.S. ISPs at 2005 year end included America Online (AOL) at approximately 18.6 million registered dial-up users (down from 27.7 million at year end 2001), for a market share of 20.1 percent. Comcast reported cable modem customers at 9.0 million, AT&T (includes SBC) at 5.7 million DSL customers, Road Runner at 5.4 million cable broadband users, EarthLink at 5.3 million through a variety of access methods, and BellSouth DSL subscribers at 3.1 million [15].

According to the Federal Communications Commission (FCC), through December 31, 2005, there were approximately 50.2 million subscribers in the United States with high-speed Internet access, which it defines as service providing speed of at least 200 kbps in at least one direction. That is an increase of 33 percent over the 37.9 million subscribers at the end of 2004. By comparison, there were 9.6 million at the end of June 2001. Cable modem service accounted for 57.5 percent of the 50.2 million lines, Asymmetric DSL (ADSL) for 40.5 percent, Symmetric DSL (SDSL) or traditional wireline connections for 0.3 percent, fiber optics for 0.5 percent, and other technologies for 1.2 percent. The FCC also reports statistics on advanced services lines, which it defines as providing speeds of at least 200 kbps in both directions. According to the FCC, there were 42.8 million advanced services lines as of December 31, 2005, an increase of 48 percent over the previous year [16].

Terminal access via dial-up connection clearly is the most common means of Internet access, certainly for the residential, or consumer, and SOHO (Small Office/ Home Office) markets, worldwide. Whether through an analog modem or an Inte-grated Services Digital Network (ISDN) modem, dial-up access makes use of the PSTN, which, as is discussed elsewhere in this book, is a circuit-switched, Time Division Multiplexing (TDM)–based network.

Direct connections to the Internet, in this context, are considered to be those that connect to a regional or national backbone provider (i.e., level 1, 2, or 3), bypassing a local ISP (level 4). Dedicated access arrangements are cost effective for large user organizations and for those who use the Internet intensively. Direct access can be on the basis of any number of alternatives, including Dataphone Digital Service (DDS) and T/E-carrier. Direct SONET access also is offered at speeds up to 155 Mbps (OC-3) by select providers. Such direct connections commonly are in the form of leased lines provided by the ILEC, although a facilities-based data CLEC also might provide such access.

· The Internet Research Task Force (IRTF) deals with long-term issues. The work of the IRTF is accomplished in small, focused research groups that work on topics related to Internet protocols, applications, architecture, and technology.

· The Internet Corporation for Assigned Names and Numbers (ICANN) is a not-for-profit organization formed in 1999 to assume the responsibilities from the federally funded Internet Assigned Numbers Authority (IANA) for assigning parameters for Internet protocols, managing the IP address space, assigning domain names, and managing root server functions. Internet protocol parameters managed by ICANN include the assignment of TCP ports, which are logical points of connection in the context of TCP, which is a part of the TCP/IP protocol suite developed for what is now known as the Internet. Port numbers provide a mechanism by which IP host computers can multiplex multiple types of concurrent connections at a single IP address. In combination, the IP address and the port number identify a socket, with a source socket and destination socket defining a TCP connection. Port numbers are 16-bit values that range from 0 to 65,536:

· Well-known ports are numbered 0-1023 for the use of system (root) processes or by programs executed by privileged users. Examples of well-known ports include 25 for Simple Mail Transfer Protocol (SMTP), 80 for HyperText Transport Protocol (HTTP), and 107 for Remote Telnet Service. In the TCP/ IP-based client/server environment, the server assigns the ports in consider-ation of the application-level protocol exercised at the client level.

· Registered ports, which are registered by IANA as a convenience to the com-munity, can be used by ordinary user processes or programs on most systems and can be executed by ordinary users. Falling in the range 1024–49151, these same port assignments are used with the User Datagram Protocol (UDP) to the extent possible.

· Dynamic ports and/or private ports are those from 49,152 through 65,535.

13.5 IP ADDRESSING

The vast majority of networks currently make use of Internet Protocol version 4 (IPv4), although IPv6 is available and used in some large, recently deployed networks. The specifics of IPv4 and IPv6 are discussed in some detail later in this chapter. Now, I want to focus attention on the specifics of the addressing scheme, with emphasis on IPv4, which was defined in RFC 791.

· Class A addresses are identified by a beginning 0 bit. The next 7 bits identify the specific network, with only a possible 128 (27). The remaining 24 bits identify the specific host computer on the network, with as many as 16,777,216 (224) possible machines. Class A addresses are intended for very large networks supporting a great number of host computers.

· Class B addresses are identified by a beginning set of 2 bits in a 10 sequence. The next 14 bits identify the network, with 16,384 (214) possible networks. The remaining 16 bits identify the specific host computer, with as many as 65,536 (216) possible machines.

· Class C addresses are identified by a beginning set of 3 bits in the binary sequence 110. The next 21 bits identify the network, with 2,097,152 possible networks. The remaining 8 bits identify the specific host computer on the network, with as many as 256 (28) possible machines. Most organizations hold Class C addresses.

· Class D addresses are identified by a beginning set of 4 bits in the binary sequence 1110. Class D addresses are intended for multicast purposes, with the remaining 28 bits specifying the multicast address.

· Class E addresses are identified by a beginning set of 4 bits in the binary sequence 1111. Class E addresses are reserved for experimental use [17].

Dotted decimal notation is the manner in which all IP addresses are written. Each 32-bit address field is divided into four fields, expressed as xx.xx.xx.xx, with each field given a decimal number value of 0 − 255, the range expressed by a single octet in binary form (28 = 256, or 0 − 255). Class A addresses begin with 1 − 127, Class B with 128 − 191, and Class C with 192 − 223. As Connect Northwest, my ISP, has an address of 206.40.133.20, it is easily identified as holding a Class C address.

There are both public and private IP address spaces. Computers on private Local Area Networks (LANs) running the TCP/IP protocol suite do not require public addresses, at least for internal use within the LAN domain. Private IP addresses are set aside and will never be used publicly. Also known as Network 10 addresses, in reference to the first field in the first address range, private IP addresses fall into the following ranges:

· 10.0.0.0 to 10.255.255.255

· 172.16.0.0 to 172.31.255.255

· 192.168.0.0 to 192.168.255.255

By convention, routers are not supposed to forward any packets to the public Internet from IP addresses in those private ranges. When a LAN-attached computer requires access to the public Internet, it is necessary that the private IP address be translated into a public IP address. The process takes place in a router that interfaces to both domains and runs the Network Address Translation (NAT) protocol, as defined in RFC 3022, which obsoleted RFC 1631. NAT works beautifully in most situations but falls short where the protocol and application demand end-to-end connectivity. In IPsec (IP security), for example, the packet headers are digitally signed and must remain unmodified from source to destination.

As noted earlier, IPv4 provides an address field size of 32 bits, which yields the potential for 232, or 4,294,967,296, distinct addresses, if it were not for the reserved address spaces. That is not quite enough to address every man, woman, and child on the face of the earth, but it is close, give or take a billion or so. And it seemed adequate for many years. The popularity of TCP/IP, especially given the commercialization of the IP-based Internet, has placed a good deal of strain on the IPv4 numbering scheme, much as the popularity of networked fax machines, cell phones, pagers, computer modems, and even copy machines has strained the E.164 numbering scheme used in the PSTN. Adding to the problem is the fact that the organization of the address space into classes, some of which are reserved for future use, wastes a lot of addresses. Further, very substantial Class A and B address spaces were parceled out to large organizations that really did not need them. As only a small percentage of the addresses were actually used, a huge number of available addresses were wasted. This is not unlike the wasteful way in which telephone numbers were until fairly recently parceled out to U.S. LECs—in CO prefix blocks of 10,000 at a time, regardless of whether they actually needed 1, or 10, or 100, or 10,000.

IPv6 resolves this issue through the expansion of the address field to 128 bits, thereby yielding 2128 potential addresses. This is a total address potential of 340,282, 366,920,938,463,463,374,607,431,768,211,456. According to Christian Huitema, that is enough to assign 32 addresses per square inch of dry land on the earth's surface, which should just about do the trick [18]. Given the proposals for assigning IP addresses to network cellular phones, Personal Digital Assistants (PDAs), coffee pots, refrigerators, heating and air conditioning systems, automobiles, and virtually every imaginable device, IPv6 clearly adds value. IPv6 also provides additional functionality, although CPE upgrades are required.

The original gTLDs are unsponsored, meaning that they operate under policies established by the global Internet community, directly through the ICANN process. The original gTLDs, their intended use, and the operators responsible for their administration are as follows:

· .arpa—address routing and parameter area, exclusively for Internet infrastructure purposes (IANA, under guidance of the IAB)

· .com—commercial organizations (VeriSign Global Registry Services)

· .edu—accredited degree-granting educational institutions (Educause)

· .gov—U.S. government agencies (U.S. General Services Administration)

· .int—organizations formed under international treaties between governments (IANA.int Domain Registry)

· .mil—U.S. military (U.S. Department of Defense Network Information Center)

· .net—net work access providers originally, now unrestricted (VeriSign Global Registry Services)

· .org—noncommercial organizations originally, now unrestricted (Public Interest Registry)

A sponsored TLD is a specialized TLD that has a sponsor representing the narrower community that is most affected by the TLD. A sponsor is an organization to which ICANN delegates some defined ongoing policy formulation authority regarding the manner in which a particular sponsored TLD is operated for the benefit of a defined group of stakeholders. Sponsored TLDs include the following:

· .aero—One of the original sponsored TLDs selected in 2000 and created officially in 2002, .aero is dedicated to aero nautical interests and is sponsored by Societe Internationale de Telecommunications Aeronautiques (SITA).

· .cat—In September 2005, ICANN and Fundació puntCat entered into a Sponsored TLD Registry Agreement under which that organization sponsors the .cat domain, which is intended to support the Catalan linguistic and cultural community. This agreement is most unusual, as Catalan is spoken by less than 16 million people in the world and understood by less than 21 million. Catalan is the language of Catalonia in Spain, the city of Valencia, the Principality of Andorra, and other isolated cities, regions, and islands in Spain, France, Italy, and that general area of Europe.

· .jobs—In April 2005, .jobs was created through an alliance between Employ Media, the Society for Human Resource Management (SHRM), Verisign, and ICANN. The .jobs TLD is specific for companies that wish to recruit employees and for job seekers and companies seeking employees. It is not now permissible for third parties (i.e., corporate recruiters) to use .jobs.

· .coop—One of the original sponsored TLDs created in 2002, .coop is dedicated to the use of cooperative associations and is sponsored by Dot Cooperation.

· .mobi—In July 2005, ICANN approved .mobi for the use of consumers and providers of mobile products and services. The TLD is sponsored by mTLD Top Level Domain and became available for registrations in May 2006. This TLD is highly controversial, as it partitions the Web along nontraditional lines of mobility and nonmobility.

· .travel—In May 2005, ICANN approved .travel as a sponsored domain for those whose primary area of activity is in the travel industry. The TLD is sponsored by Tralliance Corporation.

· .xxx—In May 2006, ICANN's board of directors voted against a proposed agreement for an .xxx domain, which would have been a TLD for pornography. Some argued that .xxx would serve as a positive, if voluntary, means of segmenting the Internet. ICANN tentatively approved the new TLD before receiving an unprecedented level of correspondence in opposition [19].

Top-level-domains also identify the country, in the form of a neutral two-character country code, as established and maintained by the ISO 3166 Maintenance Agency. Country codes, the management of which is delegated to the government of each nation, are appended to the standard address and are necessary only if the target country domain differs from the country domain of origin. The following are examples of the 241 current country codeTop-Level Domains (ccTLDs) and the countries they designate:

· .am—Armenia

· .au—Australia

· .ca—Canada

· .fm—Federated States of Micronesia

· .jp—Japan

· .nz—New Zealand

· .sw—Sweden

· .tv—Tuvalu

· .us—United States

· .za—South Africa

In March 2005, ICANN approved .eu as a regional country code for the European Union. The TLD is administered by EURid, a consortium of the ccTLD registry operators of Belgium, the Czech Republic, Sweden, and Italy. There currently are a number of other proposed regional ccTLDs under active consideration, including the regional TLDs of .asia and .africa.

IANA is responsible for administering the management of the IP address space. End users get their IP addresses from an ISP, which gets IP address allocations from a Local Internet Registry (LIR), which gets allocations from a National Internet Registry (NIR) or Regional Internet Registry (RIR). There currently are five RIRs, as follows:

· African Network Information Center (AfriNIC)

· Asia Pacific Network Information Center (APNIC)

· American Registry for Internet Numbers (ARIN)

· Latin-American and Caribbean Network Information Center (LACNIC)

· Réseaux IP Européens Network Coordination Center (RIPE NCC)

In addition to general number administration, each RIR is responsible for maintaining one or more of the root servers, maintaining a Whois database for IP ownership lookups, deployment of a routing database, coordination of ENUM delegations, and network measurement and statistical reporting.

ICANN assumed the responsibility of administering Internet addresses from IANA. Under IANA administration, the responsibility for TLD assignment was contracted to the Network Information Center (NIC) Internet Registry, commonly known as InterNIC. Now a nonprofit company operating under IANA, for a number of years InterNIC was a commercial enterprise of Network Solutions, now a VeriSign company. For the first few decades of the Internet, domain assignments were free for the asking. InterNIC then began to charge for .com domains, at the rate of $ 70 for the first two years and $ 35 for each one-year renewal. In 1999, InterNIC lost its monopoly over domain assignment, as four competing entities were approved in April 1999 for a testbed period to extend through June 25, 1999. Operational responsibility for each TLD now is assigned to an operator chosen through a more or less competitive bidding process, and there now are many hundreds of accredited registrars working under the supervision of the operators. Coordination of domain name assignment is accomplished through the Shared Registry System (SRS), so that duplicate names are not assigned. Currently, all domain names are maintained in mirrored databases on 13 root (i.e., centralized primary source) Domain Name Servers (DNSs) distributed around the world. Thousands of Domain Name Resolvers (DNRs) located strategically with ISPs and institutional networks periodically download database updates from the root servers. Through this network of resolvers, translations can be made from domain names and Uniform Resource Locators (URLs) into IP addresses, and vice versa. I discuss URLs in more detail later in this chapter.

The Internet works on the basis of IP addresses, as already discussed in this chapter. You certainly can access another host computer on the basis of an IP address, but IP addresses are hard to remember and hard to type correctly. Domain names are much easier to remember and enter, but a translation must take place to convert domain names to IP addresses; the various routers and switches depend on this in order to route the data to the correct destination device. This process of translation takes place through the Address Resolution Protocol (ARP) and through an address resolution database that resides on a server that is accessed by the originating router. As I mentioned earlier in the chapter, all domain names are maintained in mirrored databases on 13 root (i.e., centralized primary source) Domain Name Servers (DNSs) distributed around the world. Thousands of Domain Name Resolv-ers (DNRs), located strategically with ISPs and institutional networks, periodically download database updates from the root servers. Through this network of resolv-ers, translations can be made from domain names and Uniform Resource Locators (URLs) into IP addresses, and vice versa.

IP, the basic building block of the Internet, is a Layer 3 (Network) internetworking protocol for the routing of datagrams through gateways connecting networks and subnetworks. Defined in RFC 791, IP is a connectionless protocol, as no true connection is established between the source and destination devices. Rather, the IP packets are presented to the network by the originating device and handled through the network with no advance knowledge of either the existence or the availability of the destination device. IP can be characterized as datagram oriented because each IP packet works its way through the network independently, with no thought of an individual packet belonging to a larger stream of packets. IP also can be characterized as a best effort protocol, as it offers no guarantees of delivery, no sequencing, and no error detection and correction mechanism.

IP provides for packet segmentation and reassembly and provides specific addressing conventions in the form of dotted decimal notation, as previously described. IP supports routing control as well as status translation and communications. While IP has no concept of the specific content of the packet or of its service requirements, it also supports multiple service types, including low-delay, high-bandwidth, and high-reliability paths. Dial-up IP access protocols include SLIP and PPP.

The total size of the IP datagram, including the IP header, can be up to 65,535 octets in length. At a minimum, all networks must support a packet of at least 576 octets. As illustrated in Figure 13.3, the minimum size of the IP header is 20 octets. Note that the IP datagram is viewed in terms of a 32-bit width, as the original processors that implemented the IPv4 protocol had 32-bit word (i.e., value) lengths [17].

The IPv4 datagram contains the following field:

· VERS: Four bits identifying the IP version number. The version number is 4.

· IHL: Four bits of Internet Header Length (IHL). The minimum value is five 32-bit words, or 20 octets. The IHL also provides a measurement of where the TCP header, or other higher layer, header begins.

· Total Length: Sixteen bits describing the total length of the datagram, including the IP header. The maximum size is 65,535 octets (216-1, with 0 not considered as it has no value), and all network hosts must be able to handle a datagram of at least 576 octets.

· Identification: Sixteen bits that are used in fragmentation control. In the event that the receiving network cannot accommodate a datagram of the specified total length, that datagram must be fragmented. Each fragment must contain a copy of the identification field and certain other fields in the IP header so they can be reassociated and the datagram can be reconstituted.

· Flags: Three bits that define the manner in which the fragmentation occurs. The first bit always is set at 0. The second bit defines whether fragmentation is permitted. Fragmentation, for example, may not be permitted in certain applications where only the entire datagram is useful. The third bit is used to identify the last fragment in a series of fragments.

· Fragment Offset: Thirteen bits that identify where the fragment fits in the complete set of fragments that comprise the original datagram. This field is used to sequence the fragments correctly, as they may arrive at the destination device out of sequence.

· Time To Live (TTL): Eight bits that specify the length of time in seconds that the datagram can live in the Internet system. The maximum length of time is 255 seconds (28-1, with 0 not considered, as it is the official time of death), or 4.25 min. From the instant the IP datagram enters the Internet, each gateway and host that act on the datagram decrement the TTL by at least 1 s, although the time it has possession of the datagram generally is much less. When the TTL reaches 0, the datagram is declared dead and is discarded. The TTL mechanism prevents packets from wandering the Internet for eternity, at which point they would have no value and would only contribute to overall network congestion. Over time, the TTL field has been redefined to indicate, as an option, the number of hops (i.e., routers) through which the packet travels. In effect, the TTL is a hop count, anyway. The default TTL is 64.

· Protocol: Eight bits identifying the higher layer protocol that created the message contained in the data field. Examples include TCP and UDP.

· Source IP Address: Thirty-two bits containing the IP address of the source host.

· Destination IP Address: Thirty-two bits containing the IP address of the destination host.

· IP Options, If Any: An optional, variable-length field used by gateways to control fragmentation and routing options.

· Padding: A variable-length field used only when necessary to ensure that the IP header extends to an exact multiple of 32 bits.

· Data: A variable-length field that contains the actual data content [16, 17].

SLIP is the original and most basic protocol for handling IP packets in a serial bit stream across a voice-grade telephone connection. Installed on both the user's workstation and the provider's server, SLIP forwards packets created by TCP/IP software. SLIP, with origins in the 3COM UNET TCP/IP implementation from the early 1980s, is merely a Layer 2 (Link Layer) packet framing protocol that defines a sequence of characters that frame IP packets on a serial line. SLIP provides no addressing, packet-type identification, error detection/correction, or compression mechanisms. Because the protocol does so little, however, it is usually very easy to implement. SLIP is defined in RFC 1055 as a nonstandard protocol, in the formal sense, although it has become a de facto standard. RFC 1144 defines Compressed SLIP (CSLIP), a method for compressing TCP/IP performance over low-speed (300 bps-19.2 kbps) serial lines by compressing the TCP/IP headers. RFC 1144 does not deal with compressing UDP/IP headers, as they were considered at the time (February 1990) to be too infrequent to be worth the bother.

PPP performs the same basic functions as SLIP. Additionally, it performs fairly sophisticated compression in order to eliminate unused or redundant data in the headers of long sequences of packets in a transmission stream. Further, PPP supports multiple native machine and network protocols as well as subnet routing. PPP installed on a telecommuter's home PC, for example, enables communication with the home office through a router connecting to an Ethernet LAN. PPP also supports IP packet communication through the Internet [23]. RFC 1661 defines PPP. Numerous other RFCs define various PPP implementations.

The total size of the IPv6 datagram, including the IP header, is changed from that of IPv4. Specifically, datagrams greater than 65,535 octets are allowed, with these jumbo payloads identified in the payload length field. At a minimum, all network links must support a Maximum Transmission Unit (MTU) of at least 1280 octets. The IP header and any extension headers associated with the datagram are in addition to the payload length. This is a departure from IPv4, in which the size of the datagram includes any and all headers. The IPv6 datagram contains the following fields:

· Version: Four bits identifying the IP version number. The version number is 6.

· Class: Eight bits used by originating nodes and/or forwarding routers to identify and distinguish between different packet classes or priorities as set by upper layer protocols. Originally known as the priority field, this field replaces the type-of-service field in IPv4.

· Flow Label: Twenty bits used by a source host indicating any special handling requested for a flow, or sequence, of datagrams. Each datagram in the flow between an originating host and one or more destination hosts must carry the same flow label. Real-time voice, audio, or video communications are examples of applications involving flows that require non-default handling. The flow label will allow routers to identify and process Voice-over-IP (VoIP) packets more easily.

· Payload Length: Sixteen bits describing the total length of the datagram. This field is much like the total-length field in IPv4, although the payload length field does not include the IP header. As mentioned above, IPv6 supports jumbo payloads larger than the traditional 65,535 octets. At a minimum, all network links must support a Maximum Transmission Unit (MTU) of at least 1280 octets. The recommendation, however, is that network links be configured to support an MTU of 1500 octets or greater in order to support encapsulation of Ethernet payloads without incurring fragmentation. The IPv6 payload length includes any IPv6 header extensions, TCP or UDP headers, and any layer 7 headers that might be associated with the datagram.

· Hop Limit: Eight bits that specify the number of hops (i.e., routers) through which the packet can travel. Each router along the path decrements the field value by 1 until the value reaches zero, at which point the packet is discarded. This field is similar to the IPv4 TTL (Time To Live) field, with the exception that the seconds parameter has been eliminated and only the hops parameter is supported.

· Source Address: One hundred twenty-eight bits (hexadecimal) containing the IP address of the source host. (Figure 13.4 depicts this as four 32-bit rows.)

· Destination Address: One hundred twenty-eight bits (hexadecimal) containing the IP address of the destination host. (Figure 13.4 depicts this as four 32-bit rows.)

IPv6 supports multiple extension headers. RFC 1883 recommends that they be placed in the following order:

· Hop-by-Hop Options: This header carries optional information that must be examined by every node along a packet's path. This header carries information such as the type of extension header immediately following and specific instructions as to what should be done if the processing node does not recognize the option type and whether or not the option data may change en route. This header also identifies the length of the hop-by-hop header and contains padding options.

· Destinations Options: This header carries optional information that must be examined by the destination host. This header carries information such as the type of header immediately following, the length of the destination header, and padding options. Optional destination information may be contained in this header or in a separate extension header such as the fragment header or the authentication header.

· Routing: This IPv6 source node uses this header to list one or more intermediate modes to be visited along the path to the packet's destination.

· Fragment: The IPv6 source node uses this header to send a packet larger than the path MTU will accommodate. All fragmentation occurs at the source node, which is a departure from IPv4.

· Authentication: As defined in RFC 2402, this header provides a mechanism for ensuring the connectionless integrity and data origin authentication of IP datagrams as well as an anti-replay option. It might also include a nonrepudiation mechanism, which provides the origination node with confirmation of packet receipt. Included in this header may be the authentication algorithm and keys, the encryption algorithm and keys, and other security-related parameters.

The IPv6 address fields merit special discussion at this point. First, the address field of 128 bits yields 340,282,366,920,938,463,463,374,607,431,768,211,456 potential addresses compared to the IPv4 address field of 32 bits, which yields a comparatively paltry potential of only 4,294,967,296 distinct addresses. Despite the best efforts of CIDR, DHCP, and other mechanisms, IPv4 addresses clearly will be exhausted at some point in the foreseeable future, especially given the increased number of IP-addressable devices we have seen over the last few years and expect to see into the future. Other address-related enhancements associated with IPv6 include the following:

· Address Assignment: IPv6 offers much improved flexibility of address assignment through two approaches, both of which offer automatic address assignment and discovery. Stateful autoconfiguration resembles Dynamic Host Configuration Protocol (DHCP), as the configuration servers dynamically assign unique addresses to devices as they require them, drawing from a pool of such addresses. Stateless autoconfiguration employs two IP addresses and is particularly advantageous in mobile applications. One address is assigned permanently to the mobile device, and another address is used to route data to the network to which the mobile device is connected at the time. This stateless approach is much like sending a datagram to a device in care of a network and is useful in the context of mobile devices that move among pager, cellular, packet radio, wireless LAN, and other wireless networks.

· Address Types: IPv6 supports multiple address types, including unicast, multicast, and anycast (which is a new mode):

· Unicast supports communications between source-destination pairs.

· Multicast involves the communication of data to multiple hosts, each with its own IP address. Rather than the traditional approach of copying a packet stream at the originating router and then transmitting that stream to each device on a sequential, unicast basis, multicast involves a single transmission into the network. Based on its knowledge of the general physical direction in which the individual devices lie, the network fans out the packet stream to its peers, and the process continues until such time as all destination devices are located and the data are presented to them.

· Anycast is a new scheme that supports the communication between a source host and the closest member of a group of destination devices, with the group sharing a single anycast IP address. In this mode, the network routes the packet stream to the nearest device in the group sharing the address, based on the routing protocol's measure of distance. That device then assumes responsibility for forwarding the data to the group.

In total, IPv6 offers significant advantages over its predecessor IPv4. Through tunneling (i.e., the encapsulation of IPv6 packets in IPv4 packets), IPv6 is backward compatible. The real advantages of IPv6 can be realized, however, only if all nodes from source to destination are capable of running IPv6 in native mode. Therein lies the problem. For the most part, IPv6 requires a forklift upgrade. Therefore, IPv6 is finding its way fairly slowly into the domain of internetworking. While the emerging IP-based networks, both private (e.g., Internet2) and public, are building from the ground up, they easily can implement IPv6. They also must run dual protocol stacks to support both versions in native mode. Similarly, new backbone router implementations in large enterprise networks can support both versions easily. Gradually, IPv6 will supplant IPv4, as it works its way from the backbone to the desktop, but the full process may well take decades [17, 18, 24, 25]. IPv6 deployment has been slow to date for reasons including the cost and complexity of conversion from IPv4. Also, the pressure on the IPv4 address space has relaxed due to the success of DHCP and CIDR. The widespread use of private IP addresses within the enterprise and the use of NAT for address translations associated with traffic to the wider Internet also have made IPv6 less of an imperative. On the other hand, NAT can create bottlenecks and, in combination with firewalls, can even deny VoIP traffic. IPv6 can eliminate the need for NAT, at least as far as the addressing scheme is concerned, although network administrators likely will still use NAT for security purposes, but perhaps on a more selective basis.

Many analysts, including myself, see wireless communications as the main driver behind IPv6. The proliferation of IP-enabled wireless cell phones, PDAs, and other hand helds undoubtedly will continue for many years to come and will require IP addresses beyond the limits of IPv4. Additionally, and as noted above, IPv6 offers much improved security and mobility as compared to IPv4. Further, the emergence of 2.5 G and 3 G wireless networks present a ground-up opportunity for IPv6 deployment.

In the private sector, Bechtel, the global contracting company, is moving to IPv6 in order to maintain compatibility with the federal government as well as its customers in Asian countries that are deploying the protocol aggressively [27]. Cray Computer announced in April 2005 that it was to be the first to make commercial use of MCI's IPv6 services, citing the requirement for interoperability testing and development of next-generation software and applications for its X1 Supercomputer product line [28]. Despite the obvious advantages of IPv6, it has been very slow in gaining acceptance.

Transmission Control Protocol (TCP) is a Layer 4 (Transport) protocol defined in RFC 793. TCP evolved from ARPANET's Network Control Protocol (NCP), which was developed to provide reliable transmission across the essentially unreliable media of analog UTP and packet radio (e.g., AlohaNET). In support of higher layer applications, TCP can be characterized as making use of virtual circuits in support of byte-stream-oriented communications. As a connection-oriented protocol, TCP supports status exchange and synchronization over virtual circuits. TCP provides for file segmentation into packets on the transmit side and for reassembly on the receiving end. TCP also provides for packet sequencing, end-to-end flow control, and error control [29, 30], thereby guaranteeing delivery. Each packet in a stream of packets received by the destination device is either acknowledged as having been received correctly or requested to be retransmitted in the event of corruption. Packets that are not acknowledged or requested for retransmission are considered unacknowledged and are retransmitted by the source host.

The TCP unit of data transfer between two host computers is known as a segment. Segments are used to establish connections, transfer actual data, acknowledge packet receipt and request retransmissions, and terminate connections. Figure 13.5 provides a view of the TCP header and its component fields. The standard size of the TCP header is 20 octets, although 4 additional octets may be used to accommodate options.

The TCP header fields are defined as follows:

· Source Port: Sixteen bits that define the TCP port number used by the source application program. As discussed earlier in this chapter, TCP ports are logical points of connection. Well-known ports are numbered 0-1023 for the use of system (root) processes or by programs executed by privileged users. Examples of well-known ports include 25 for SMTP (Simple Mail Transfer Protocol), 80 for HTTP (HyperText Transport Protocol), 107 for Remote TELNET Service, and 110 for POP3 (Post Office Protocol version 3).

· Sequence Number: Thirty-two bits that identify the position of the data in the TCP segment relative to the entire originating byte stream. A fundamental notion of TCP is that each octet in a byte stream is numbered. The sequence number refers to the number of the first octet in a given segment. This field is critical if data are to be sequenced properly at the destination host. It also is critical in order that positive acknowledgments can be sent to the originating host and retransmissions of segments can be requested.

· Acknowledgment Number: Thirty-two bits that identify the acknowledgment number of the octet that the source expects to receive next. The acknowledgment number explicitly acknowledges that all previous data octets associated with all previous segments were received correctly.

· HLEN: Four bits that specify the segment header length, in 32-bit multiples.

· Reserved: Six bits reserved for future use.

· Code Bits: Six bits that define the purpose and contents of the segment. Examples include acknowledgment, connection reset, and end of byte stream.

· Window: Sixteen bits that advertise the size of the sender's sliding receive window (i.e., how much data the host computer is willing to accept, based on buffer size). A window of zero indicates that the receiver is overwhelmed and can not accept any further data until further notice. Large windows indicate that as many as 65,536 unacknowledged bytes can be in transit at a given time. Congestion, which is indicated by expiration of the retransmission timer without an acknowledgment, reduces the window size by half, thereby slowing the transmission rate.

· Checksum: Sixteen bits used for error control in the data field as well as the header. The process is that of cyclic redundancy check.

· Urgent Pointer: Sixteen bits that identify urgent out-of-band data (i.e., data not part of the information stream). Such data are treated on a high-priority basis, in advance of data stream octets that might be awaiting consumption by the destination hosts. Urgent data, for example, might include a keyboard sequence to interrupt or abort a program.

· Options, If Any: Twenty-four bits that address a variety of options, such as Maximum Segment Size (MSS).

· Padding: Eight bits in an optional field used only when necessary to ensure that the TCP header extends to an exact multiple of 32 bits. This field is used only when the options, if any field is used.

As defined in RFC 768, User Datagram Protocol (UDP) is a Layer 4 (transport) host-to-host protocol that is much simpli fied in comparison to TCP. Historically used to send datagrams between application programs, UDP offers the same unreliable, connectionless datagram delivery as IP. Like TCP, UDP uses IP for addressing and routing purposes. Unlike TCP, UDP provides no sequencing, error control, or flow control mechanisms. An application program that uses UDP assumes full responsibility for all issues of reliability, including data loss, data integrity, packet latency, data sequencing, and loss of connectivity. UDP is used extensively in VoIP and stream-oriented multimedia applications, where compression techniques are designed to mitigate such issues over a highly shared packet network. UDP also works well where transactions are of such short duration that connection setup overhead comprises a large proportion of overall transaction traffic, with DNS and SNMP exchanges being good examples.

The standard size of the UDP header is eight octets; it comprises the following fields, as illustrated in Figure 13.6:

· Source Port: Sixteen bits that define the UDP port number used by the source application program.

· Destination Port Sixteen bits that define the UDP port number used by the destination application program.

· UDP Message Length: Sixteen bits that identify the length of the message in the data field.

· Header Checksum: Sixteen bits used for error control in the header only. This checksum need not be used; if the value is set to zero, it is disregarded. This lack of regard for header control is possible because it also is accomplished in the IP header. The checksum process is that of CRC

· Data: A variable-length field that contains the actual data content [6].

Note that the entire IPv4 datagram has a maximum length of 65,535 octets and must be fragmented to fit into an Ethernet payload if it exceeds 1500 octets. The IP header consumes 20 octets. If TCP is used, the TCP header is 20 octets. If UDP is used, the UDP header is 8 octets. The balance is available for application layer data subject to any limitations of the local network.

Application-layer protocols (Layer 7 of the OSI Reference Model) in the IP protocol suite function above TCP/IP in support of specific Internet applications. Examples include the following.

TELecommunications NETwork (TELNET) is perhaps the oldest Internet application protocol. Defined in RFC 854 (May 1983), TELNET provides terminal emulation over a TCP connection, enabling the user to assume control over the applications that reside on a remote system. Virtual network terminal services permit the DTE to emulate other terminal devices, transparently, in a client/server environment.

File Transfer Protocol (FTP) supports the exchange of files between two hosts across the Internet. Defined in RFC 959, FTP also supports interactive user interface in which humans must interact with a remote host. The specifics of the file type and format [e.g., ASCII, EBCDIC, or binary and compressed or uncompressed] of data can be determined from client to server. FTP also requires clients to satisfy security authorization in the form of login and password. FTP makes use of TELNET for control messages between the hosts and relies on connection-oriented TCP for data transfer.

SNMP supports the exchange of network management information between hosts, typically including one or more centralized network management consoles that manage larger numbers of network elements in real time. Defined in RFC 1157, SNMP operates over UDP, thereby avoiding the overhead associated with TCP. There are three versions: SNMPv1, v2, and v3. (Note: A host is a computer that runs an application program. Hosts include hand-held computers, such as PDAs, as well as more substantial machines, such as desktops, servers, and mainframes.)

All of the protocols discussed so far in the context of the IP protocol suite have been data oriented, but that is by no means the end of the story. The Internet and other IP-based networks increasingly are being used to support real-time voice, audio, and video applications, all of which are extremely demanding in terms of latency, jitter, and packet loss. The Internet and its original underlying protocols were never intended to support Quality of Service (QoS), which is exactly what each of these traffic types requires. As is discussed earlier in the chapter, the IPv4 header contains a Service-Type, or Type-of-Service (ToS), field that is used to request QoS in terms of parameters such as packet precedence (i.e., priority), low delay, high throughput, and high reliability. (Refer back to Figure 13.3.) But that is about as far as it goes, and there certainly are no guarantees. IPv6 goes a bit further (pun intended) with its class and flow label fields. Originating nodes and/or forwarding routers use the IPv6 class field, which replaces the IPv4 ToS field, to identify and distinguish between different packet classes or priorities, as set by upper layer protocols. The flow label field is used by a source host indicating any special handling requested for a flow, or sequence, of datagrams, with each datagram in the flow between an originating host and one or more destination hosts carrying the same flow label. Real-time voice, audio, or video communications are examples of applications involving flows that require nondefault handling. The IETF also has developed a number of protocols that are required to support real-time and multimedia applications more effectively. While they do not provide ATM-like QoS guarantees, they can come reasonably close under the right circumstances. The next few sections examine these protocols, which include DiffServ, MPLS, RTP, RTCP, RSVP, and RTSP.

The IETF defined the Resource Reser-Vation Protocol (RSVP) in its RFC 2205 (1997), which it updated in RFC 2750 (2000). RSVP is a Layer 4 (Transport Layer) control protocol that operates on top of IPv4 or IPv6 and depends on existing and future routing protocols. As a control protocol, RSVP does not transport data. Rather, RSVP operates on a hop-by-hop basis in order to signal QoS requirements for unicast and multicast data flows to each node and, thereby, reserve the necessary per-session resources from end to end across an IP network. RSVP can operate in conjunction with other QoS protocols, including DiffServ and MPLS, to effect service discrimination [31–34].

The Real-Time Streaming Protocol (RTSP) is defined in RFC 2326 as an application-level protocol for control over the delivery of data with real-time properties, such as audio and video and including both live data feeds and stored clips. RTSP is intended to control multiple data sessions, provide a means for choosing delivery channels such as UDP, multicast UDP and TCP, and provide a means for choosing delivery mechanisms based upon RTP. There is no notion of an RTSP connection. Rather, an RTSP server maintains a session labeled by an identifier.

RTSP establishes and controls either a single or several time-synchronized streams of continuous media such as audio and video. It does not typically deliver the continuous streams itself, although interleaving of the continuous media stream with the control stream (e.g., RTCP) is possible. In other words, RTSP acts as a network remote control for multimedia servers [33, 35].

MultiProtocol Label Switching (MPLS) is defined by the IETF in RFC 2702 as a label-swapping framework with Network Layer (Layer 3) routing. Integrating Layer 2 (Data Link Layer) information about network links into Layer 3 routing logic, MPLS is based on several vendor-specific protocols, including Cisco's Tag Switching, Ipsilon's IP Switching, and IBM's ARIS technology. MPLS is designed to work through routers at even higher speed than ATM switches, while realizing much of the flexibility of an IP-based network. MPLS enables routers to make packet-forwarding decisions very quickly on the basis of short labels, comparable to Frame Relay Data Link Connection Identifiers (DLCIs), rather than making complex routing decisions after analyzing lengthy packet headers.

MPLS works on the basis of Forwarding Equivalence Classes (FECs) and flows. A flow consists of packets between common end-points identified by features such as network addresses and port numbers. An FEC is a class of packets, all of which are treated the same in terms of destination, priority level, and so on. As a flow begins, the first packet exits the user's client workstation, traverses the LAN, goes through an enterprise router, and reaches the ingress Label Edge Router (LER) at the edge of the carrier network. The LER identifies the flow based on the IP header, the interface through which the packet arrives, the packet type (e.g., unicast, multicast, or anycast), or perhaps information in the Type of Service (ToS) field.

From edge to edge through the core of the network, each LSR makes note of the incoming port number and analyzes the label associated with each packet in order to select the appropriate LSP over which the packet is to be forwarded on its way to the next LSR. Thereby, and through a series of links, the end-to-end path is set up and maintained for a given traffic flow. The more complex processes of complete header analysis and routing table lookup are performed only at the ingress edge of the network. In the core of the network, only the abbreviated MPLS label is analyzed in order to make a relatively simple and straightforward packet-forwarding decision. All in all, the routing process is simplified and latency is reduced. At the egress LER, the tag is removed, as it is no longer needed [40–43].

The structure of the 32-bit MPLS header is as follows and as illustrated in Figure 13.9:

· Experimental (EXP): The EXP field of 3 bits is used to indicate the precedence, or packet-queuing priority, for CoS purposes.

· Stack (S): The stacking bit is set at 1 to indicate the last (i.e., innermost) MPLS header in a stack of headers. Outer tags carry a 0 bit in this position. MPLS VPNs involve hierarchical routing logic that requires multiple headers. As many as four MPLS headers can be contained within a stack.

· Time To Live (TTL): The TTL field is copied from the IP TTL. The TTL is a hop count, with a default value of 64.

MPLS addresses congestion issues in a much different way than a typical enterprise Wide Area Network (WAN), in which IP routers use OSPF or some similar routing protocol to interconnect Frame Relay or ATM Permanent Virtual Circuits (PVCs). This approach invariably leads to some level of congestion, which contributes to latency, jitter, and even loss. MPLS supports constraint-based routing that considers factors such as bandwidth, hop count, and performance requirements of the traffic flow in selecting the LSP. MPLS also enables traffic engineering, as it allows the carrier to select predetermined LSPs along which the traffic will transit. In a well-designed carrier network, traffic engineering ensures that the performance requirements of the traffic flows are satisfied. MPLS-based traffic engineering also supports backup LSPs that can be invoked quickly in the event of the failure of a physical link or node.

Through its traffic engineering attribute, MPLS supports the consolidation of multiple traffic types through a single network. However, each traffic type can be associated with a Class of Service (CoS). The carrier can take one of several approaches to honoring the CoS. One approach involves a single LSP between the ingress and egress LSRs. Traffic that flows on that LSP queues in the LSRs in consideration of the precedence bits in the MPLS header, with the high-precedence packets going to the head of the queue and the low-precedence packets remaining in the rear of the queue. Another approach involves multiple LSPs, each of which is engineered to perform according to a set of parameters that considers the performance requirements of a given traffic flow. So real-time voice and video might take one LSP, business-critical data might take another, and best effort e-mail might take yet another. MPLS CoSs are definable by the carrier on a standard and custom basis. AT&T, for example, defines the following standard CoS types:

· C2: Interactive CoS designed for critical applications requiring priority treatment. Examples include one-way video broadcast systems requiring delivery confirmation within a short period of time or database activity.

· C3: Enhanced CoS designed for low latency and loss. This CoS is intended for non-mission-critical, yet delay-sensitive, applications such as SNA or streaming video.

· C4: Standard CoS designed for bursty traffic with minimal latency requirements. Example applications include e-mail and Internet traffic

CoS is not at the same level as ATMs' guaranteed Quality of Service (QoS), of course, but it can come reasonably close in a well-engineered network. Consider the following analogy. When you send a personal letter or parcel through the mail via the U.S. Postal Service, you fill out the complete address by longhand or perhaps by typing out a label that you adhere to the letter or parcel. That full address contains the name of the intended recipient, the street address and apartment or suite number, the city, the state, and the ZIP code or other postal code. You also identify whether the letter is to be sent as bulk mail, first-class mail, Priority Mail, or Express Mail. As the letter is first processed at the edge of the system, a computer system scans the ZIP code through the use of Optical Character Recognition (OCR) or Optical Mark Recognition (OMR), and a postal code is added in the form of a bar code. In the event that the OCR or OMR software cannot translate the handwritten ZIP code, a postal worker adds the bar code manually. As the letter works its way through the core of the system, electronic devices rely solely on the bar-coded ZIP code to forward the letter to the destination edge of the postal processing network. All mail destined for the same local post office is batched and delivered there. At that office, the street address is read in full and the mail is sorted for individual routes and carriers. The carrier verifies the address before depositing the mail in your mailbox. Consider the bar-coded ZIP code as analogous to a tag—it adds additional overhead but speeds up the process of sorting and forwarding in the core of the network of mail-processing centers.

There also are different paths and different levels of precedence for different types of mail. Overnight mail takes an express route by air, taking precedence over priority mail, which also travels by air. First-class mail travels by a surface mode, such as truck and ship, taking precedence over media mail and parcel post, which travel on a space available basis. Such is the case with MPLS.

MPLS works with DiffServ and RSVP and with Frame Relay, ATM, and Ethernet and is used in support of IP-based Virtual Private Networks (VPNs). MPLS works with various routing protocols such as OSPF, RIP, and BGP that can be used to distribute the routing labels. Hence the use of the word multiprotocol [33, 44, 45]. MPLS currently operates on only a single domain, as there is no Network-to-Network Interface (NNI) specified for interoperation between carriers. Any such interoperation, therefore, must be by special arrangement.

As a final variation on the theme, enter Generalized MPLS (GMPLS), once known as Multiprotocol Lambda Switching but since extended to other media. GMPLS extends MPLS beyond packet-switched interfaces to include Time Division Multiplexing (TDM), Wavelength Division Multiplexing (DWDM), and Add/Drop Multiplexing (ADM). GMPLS adds the concept of label switching to photonics at the lambda (i.e., wavelength) level in a DWDM system, to time slots in a SONET/SDH system, and to physical fibers in an optical cross-connect or photonic cross-connect. The goal of GMPLS is to speed the provisioning of end-to-end traffic-engineered paths in the TDM and optical domains, much as MPLS has done in the IP domain [40, 46].

13.8 INTERNET APPLICATIONS

Internet applications are growing as fast as the imagination and technology will support. These applications include e-mail, file transfer, Bulletin Board Systems (BBSs), library catalogs, online banking, video and radio broadcasting, and even voice telephony.

Electronic mail (e-mail) clearly is the most popular application. Somewhere in the neighborhood of 84 billion e-mail messages comprising trillions of bytes transverse the Internet every single day. E-mail transport through the Internet makes use of the Simple Mail Transfer Protocol (SMTP). E-mail messages commonly are sent with MIME attachments, which may be in the form of compressed images, video clips, and audio clips.

A Bulletin Board System (BBS) is a computer system running software that allows one to connect to what is essentially an electronic bulletin board. BBSs originally were quite local phenomena, as access was on a dial-up basis and only a local call is toll free. BBSs typically were, and often still are, run as a hobby or side interest of the sponsor. Generally, anyone can access the BBS to post messages, reply to messages, post software applications for downloading by others, play games, and otherwise communicate and share with others. BBSs were quite the rage in some Internet circles of interest until the mid-1990s but generally are considered somewhat primitive in the context of the modern World Wide Web (WWW). They remain quite useful, however, for special interest groups such as fans of the Grateful Dead rock band, system administrators struggling with software applications in a specific network environment, or attempting to deal with unusual technical issues.

Library catalogs for hundreds of libraries are available through the Internet. Such catalogs include the U.S. Library of Congress, the Research Libraries Information Network, and many major colleges and universities. Through the Interlibrary Loan Program, Internet users can have the document loaned to their local library.

Real-time applications are growing at significant rates. There are a number of highly legitimate applications and many that generally are considered to be less so. Real-time applications include collaborative design and development, interactive role playing, interactive remote education, and chat lines, which operate like simplified real-time BBSs. Other applications include voice and videoconferencing, network games, gambling, and streaming audio and video broadcasting (cybercasts).

Online banking has received a good deal of interest. While security issues abound, there is little doubt that this application has gained a broad level of acceptance. In the area of financial services, online stock trading has made a huge impact, having taken significant market share away from conventional brokerage houses. The downside of online stock trading is that several of the Internet-based brokerages have experienced numerous short-term system failures that have resulted in the total loss of their ability to execute trades on behalf of their customers.

Radio broadcasting makes use of packet audio systems such as RealAudio from RealNetworks. Although there always are issues of latency and loss in a real-time stream-oriented application, AM quality is quite possible. In other words, the general quality is not necessarily great, but the price is right. Some years ago, a number of students from Asia were enrolled in a graduate-level course I taught at the University of San Francisco. Several students from Japan, in particular, regularly listened to radio broadcasts from Tokyo over the Internet. Again, the quality was not great, but the price was right (read free). Additionally, they really had no other means of hearing such a broadcast or getting information from home on a timely basis. My wife, Margaret, also a consultant, several years ago had a long-term engagement with a large wireless company. For the first few weeks, she complained about the fact that the office was as quiet as a tomb. Not only was there none of the usual water cooler conversation, but people seldom ventured out of their cubicles. One evening when we were shopping at the local office supply store, she stopped to buy an expensive headset, which puzzled me. It seemed as though she had figured out the reason for the dead silence in the office. Margaret was about to join the other hundred or so staffers who listened to radio broadcasts on the Internet while working. Bandwidth and quality were not particular problems because the company had several T3 circuits connected directly to the Internet backbone.

The quality of a VoIP call over the Internet varies. My lovely bride, Margaret, is the principal of The Evergreen Group, a curriculum company that develops training material for companies all over the world. She recently managed a project involving instructional designers and subject matter experts in Argentina, Australia, Canada, Columbia, New Zealand, Norway, Spain, and the United States. In order to keep long-distance calling costs to a minimum, she used a popular Internet voice service. There were two ways to connect a call:

· Softphone to Softphone: She could call from her softphone (i.e., computer equipped with the VoIP software) to the other person's softphone through the service provider's website. In this scenario, the call was entirely over the Internet, and both parties were required to have broadband, always-on access. The quality was sometimes quite acceptable and sometimes awful due to latency, jitter, and loss. The call was free, but usually not free enough.

· Softphone to PSTN: She could call from her softphone to the other person's PSTN hardphone or cell phone. In this scenario, the international portion of the call was over the Internet to a local gateway in the distant city, where the connection to the PSTN (and cellular network) was established. The quality generally was quite good and the cost of the call compared quite favorably to the cost of a traditional PSTN call.

As you might expect, some of the traditional PSTN carriers are very upset about this application because it threatens to shift large volumes of traffic away from the conventional PSTN and to the Internet. Some carriers in the United States have even blocked VoIP calls. Many nations, particularly developing countries, have even outlawed the practice, at least for end users, as it so negatively affects the associated revenue stream, which oftentimes is a significant source of hard currency. On the other hand, many of those same nations have contracted, through their monopoly national carriers, with IP-based international carriers that use both private IP networks and the Internet.

Internet telephony actually offers more than just low cost in return for low quality. Consider a convergence scenario in which you access the Internet and the Web through your ISP to view a website to make a reservation or order a product or service. As you click your way around the website through your browser, you have a question about the specifics of a product. You click the Talk to An Agent button and are connected to a real person with whom you can carry on a conversation over the public Internet using VoIP technology. While you are talking, you both view the same information on the website. The agent can take control of the website to present you with the proper information, which saves you a lot of mouse clicks, thereby improving communications and enhancing your Internet experience. The voice quality may be less than toll quality, but the convergence of voice and data has added considerable value overall.

Now, take this experience to the next level. If you access your ISP over an ADSL local loop and connect to a private next-generation IP-based carrier, rather than through the public Internet; then your converged voice and data communications might transport over an improved IP-based backbone optimized for voice as well as data. Your voice communication will be improved. Your data communication probably will be more satisfactory as well. At this level, Internet telephony really is not Internet telephony at all, except perhaps at the very edges. Rather, the majority of your experience is one of voice and data communications over a finely tuned, IP-based network.

There currently are various proposals to impose a tiered, content-sensitive pricing arrangement for Internet usage whereby real-time, stream-oriented applications such as voice and video would be charged at a premium. These proposals are the source of considerable controversy, of course.

13.10 INTERNET2

Initially, Internet2 made use of existing networks such as the very-high-speed Backbone Network Service (vBNS), a network provided under a cooperative agreement between the NSF and Worldcom (now MCI, which is part of Verizon) in support of NSF-approved institutions of higher learning. vBNS initially ran over an ATM/SONET backbone at 155 Mbps (OC-3) and later was upgraded to 622 Mbps (OC-12). More recently, the Abilene Project was developed as a high-performance network by the University Consortium for Advanced Internet Development (UCAID) in cooperation with Qwest Communications, Nortel Networks, Cisco Systems, and Indiana University. Abilene supports the Internet2 infrastructure through high-speed routers connected to several dozen GigaPOPs (i.e., Gbps Points of Presence, routers positioned as access points) interconnected over optical fiber transmission facilities running at speeds up to 10 Gbps (SONET OC-192) nationwide [47, 48]. Abilene is now being phased out in favor of a DWDM backbone that initially will support ten 10-Gbps lambdas and eventually will scale to eighty 10-Gbps lambdas [49].

Géant2, a similarly ultrafast European backbone, was lit in Milan in June 2005, with the 186 million Euros funded half by the European Union and half by member nations. Géant2, which translates from French as Giant2, connects some 5000 institutions across Europe, including many high schools, in the European Union, plus Russia, Switzerland, Turkey, and Israel. Anyone connected with any of those institutions has access to the network [50].

13.11 WORLD WIDE WEB

The World Wide Web, also known as the Web, is the Internet's first killer application. Tim Berners-Lee developed the Web at the l'Conseil Europé en pour la Recherche Nuclé aire (CERN), which translates from French as The European Council for Nuclear Research and is generally known as the European Laboratory for Particle Physics in Geneva, Switzerland. The Web is a multiplatform operating system that supports multimedia communications on the basis of a Graphical User Interface (GUI). The GUI provides hypertext, which enables the user to click a highlighted text word and search related files across Web servers and through hot links; in other words, the Web is hyperlinked. In addition to text, the Web supports graphics, audio, and video, with various levels of quality and speed depending on the bandwidth available.

Organizations or individuals can develop a website, consisting of a computing platform (server) connected to the Web on a full-time basis. Although the typical business approach involves the deployment of a dedicated server, smaller users may achieve the same end through renting website capability from a hosting ISP. The ISP logically partitions a server in support of multiple users and multiple home pages.

Home pages effectively provide a POP on the Web for businesses, typically supporting advertising and informational purposes. Within each website is a home page, or multimedia informational document, which may contain graphics, animated graphics, video clips, and audio clips as well as text. Individuals may develop personal home pages, which may offer updates on personal life along the lines of a cyberspace version of the ever popular, What I did on my summer vacation. (Don't laugh. We've done the same thing here at the Horak household.)

Web applications are in a world of their own, including advertising, publishing, micromarketing, catalogs and direct sales, entertainment, and e-commerce. The great hope is that the Web will become a significant tool for commerce, enabling customers to access a home page on a website, gain information about a product or service, and actually make a purchase online.

Home page sponsorship is offered by several directories and frequently hit portals and home pages. Virtually all browser portals and home pages for publishers of periodicals, for example, sell sponsorships. In return for a rather princely sum, your home page banner will appear on their home page; users can click your banner and hyperlink to your home page.

Personal home pages exist by the millions, at least. Text describing collections of airsick bags, photos of the children, audio clips of the growls and screeches of family pets, and video clips of summer vacations—all are possible and all are present on the Web. The latest variation on this theme is the blog, a contraction of We b log. A blog is a website where an individual maintains a personal journal or even an interactive forum much like a personal newsgroup.

Popular in the blog community is a software application program known as a news aggregator, which maintains contact with selected sites in the blogosphere area in cyberspace that make content available in RSS format. RSS is a push technology, that is, a metadata technology, that can identify changes in data and initiate a content push to the end user without the user having to search it out and pull it from the site. The term RSS is an umbrella term variously used to describe a number of versions of several data Web feed formats specified in eXtensible Markup Language (XML) and used for syndication of Web content. Those standards include:

· Really Simple Syndication

· Rich Site Summary

· RDF Site Summary. Resource Description Framework (RDF) is a W3C specification that integrates a variety of applications using XML as an interexchange syntax.

More recently, the IETF adopted the Atom Publishing Protocol (APP), which builds on the previous RSS work but corrects some RSS deficiencies, according to some. The Atom format is documented in the Atom Syndication Format (RFC 4287).

Electronic commerce, or e-commerce, has grown to incredible proportions over the past several years. The U.S. Census Bureau defines e-commerce as "sales of goods and services where an order is placed by the buyer or price and terms are negotiated over an Internet, extranet, Electronic Data Interchange (EDI) network, electronic mail, or some other on-line system. Payment may or may not be made online." According to that definition, e-commerce retail sales in the United States reached US $ 85.993 billion in 2005, which is approximately 2.2 percent of the total retail sales of US$3.861 trillion [54].

As noted earlier in this chapter, the Internet Tax Freedom Act became law in October 1998. One of the provisions of that act protects e-commerce from taxation at the state and local levels, in effect declaring the Internet and the Web a tax-free zone in terms of sales taxes for interstate transactions. In other words, your online purchases from a company in Rhode Island are not taxed, assuming that you live outside the state. Technically speaking, no taxes apply as long as the seller does not maintain a physical presence in the buyer's state. This step is grounded in tradition, as mail-order catalog companies long have enjoyed the same advantage. For that matter, the purchases that you make in New York City are exempt from state and local sales taxes if you have them shipped to me in Mount Vernon, rather than taking them back to your hotel and hauling them back home yourself. And, as most people know, the cost of postage easily can be less than the sales taxes. Now, buyers are required to pay levies on their online and mail-order purchases when they file their state income tax returns, assuming that there is an income tax in the state of resi-dence, but it is unlikely that any but the most right-minded consumer actually would do such a thing. The risk, of course, is that the increasing popularity of e-commerce can undermine the state and local tax bases, which fund a wide variety of social and other services.

The huge number of servers and incredible amount of information available on the Internet and Web quickly made it difficult to find the desired resource. In other words, you could not find the information unless you knew pretty much where to find it—sort of like having to know at least the approximate spelling of a word in order to look it up in the dictionary to determine the exact spelling. This problem was addressed in the early 1990s through the development of search engines and, more recently, Web browsers.

Search engines are in the form of client/server software application programs that support the search for Internet informational resources through the development and maintenance of resource directories. In contemporary terminology, search engine commonly is used to describe systems such as Alta Vista, Ask Jeeves, Excite, Google, LookSmart, Lycos, Infoseek, and Yahoo! (Yet Another Hierarchical Officious Oracle).

Web browsers are software application programs used to locate, display, and interact with Web pages. The first primitive text browser appeared in 1991, courtesy of CERN. In 1993, the first graphical browsers appeared: Viola for X Windows, Mac browser from CERN, and Mosaic for X Windows [55]. Early graphical browsers that also support hyperlinks include Archie, Gopher, JUGHEAD, VERONICA, and WAIS. Currently, the most popular browsers include Microsoft's Internet Explorer, Mozilla Firefox, Apple Safari, AOL's Netscape Navigator, and Opera. Select search engines and browsers include the following:

· Gopher was developed at the University of Minnesota, where the Golden Gopher is the school mascot. Developed as a user interface to provide easy access to server resources in educational institutions, Gopher has become a de facto user interface standard. Gopher servers enable the user to access a directory of over 1800 Gopher server sites, click the name of the server, and browse its file resources on the basis of nested menus. Gopher requires that the user know the server on which the subject file is located, somewhere in gopherspace.

· Very Easy Rodent-Oriented Netwide Index to Computerized Archives (VERONICA) is an Archie variation that supports an index of gopherspace titles on which a search can be performed. The selected resources are then delivered to the user in the form of a Gopher menu.

· Jonzy's Universal Gopher Hierarchy Excavation And Display (JUGHEAD) is similar in operation to VERONICA, although it limits the search to a specific organization. JUGHEAD also delivers a custom menu of available resources located on the basis of the keyword search. Playing off Archie and Gopher, subsequent developers of search mechanisms tried to stay with the Archie comic book/rodent theme, proving once and for all that even acronyms can be fun. Alas, the more recent and more powerful browsers were not named with a noticeable sense of humor.

· Mosaic is a browser developed by a team led by Marc Andreessen at the National Center for Supercomputing (NCSA) at the University of Illinois Urbana-Champaign campus. Mosaic provides a consistent user interface available in versions to support Macintosh, Microsoft Windows, and Unix-X Windows. Mosaic can be used over dedicated or dial-up Internet connections; the dial-up access provider must support either SLIP or PPP Mosaic enables the easy browsing of Web resources through menus that support hypertext. Through a simple process of mouse clicking, the user can select menu options. Selected files can include audio and graphics, both of which can be viewed without the requirement to download the subject file. Mosaic technology is licensed by the NCSA for commercial application. Spyglass licensed well over 12 million copies of Mosaic to IBM, DEC, and others who intended to resell the company's Enhanced Mosaic, which included enhanced security mechanisms based on Secure HTTP (S-HTTP). Ultimately, Spyglass Mosaic was licensed to Microsoft, where it formed the basis for Internet Explorer.

· Wide Area Information Service (WAIS) servers enable the user to specify the databases requested for search and to conduct a subject matter search on the basis of keywords.

Earlier in this chapter, I discussed the concept of access anywhere in the context of the Internet. Our focus there was on e-mail messaging. Well, that is by no means the end of the story. Contemporary cell phones, PDAs, and other hand helds increasingly are Web enabled through built-in microbrowsers. As hand helds just do not offer the same capability as a laptop or desktop in terms of processing power and display technology, and as wireless networks just do not offer the same bandwidth or error performance as wired networks, some adjustments have to be made to support an effective Web experience. Those adjustments largely involve tailoring the content so that it fits comfortably on a smaller display that does not support colors other than black and white. Further, the complexity of the website must be adjusted in consideration of the limitations of the user interface of a hand held, which currently does not support point-and-click capability easily, as it does not support a mouse. There are two competing approaches: Internet Mode (i-Mode) and Wireless Access Protocol (WAP):

· Wireless Access Protocol is a carrier-independent, device-independent, transaction-oriented protocol employed in cellular networks outside of Japan, also in support of text, graphics, and audio. While the best performance is achieved when accessing websites written in WML, which is similar to but different from HTML, this requires that the content be rewritten. The alternative is transcoding from HTML to WML, which is accomplished through gateways. A much simpler but much less attractive technique is Web clipping, which strips the graphic content out of Web pages. Security over the wireless link is provided through Wireless Transport Layer Security (WTLS), pronounced witless

The most popular applications for wireless Web access generally are thought to include services such as stock quotes, weather forecasts, airline and train schedules, and weather reports. Location-based services, such as finding the closest restaurant or bar, are not only possible but are also highly attractive to wireless users. Wireless banking also holds great promise, with the eventual hope being that hand helds could replace cash, checks, and credit cards altogether [56–60].

13.12 INTRANETS AND EXTRANETS

Intranets are the concept of the Internet turned inward—an unexpected turn, perhaps, but a very significant one. Intranets essentially are mini-Internets deployed within organizations or groups of organizations. They can be in the form of internal Internets, functioning to provide access to information resources within the company, university, or other organization. They can be confined to a campus environment or can extend across the wide area to link together multiple, geographically dispersed locations. They also can function as a closed subnet of the Internet, much as is intended for Internet2 in the college and university market. Although conceived as recently as 1995, intranets have spread quickly to the point that most medium to large user organizations have them in place. Intranets use the same browsers as used for Internet application, thereby avoiding the training and support requirements imposed by another application software package such as groupware.

Intranets can be used for communications to and between employees for just about any purpose imaginable. A number of corporations use intranets to keep their employees advised of company policies, job postings, company events, product literature, press releases, and so on. With the proper password for security purposes, of course, privileged users can access sensitive internal company information, including customer billing records and network usage data. As is the case with the Internet and Web, images, video clips, and sound clips associated with textual information are supported. Hypertext links can be included to hot link to other sites and databases and even to the Internet and the Web.

Health care organizations have made fairly extensive use of intranets to link remote clinics, reducing paperwork and abbreviating communication time. Kaiser Permanente has put the intranet concept to use in order to keep employees abreast of changes in health legislation and insurance law as well as to provide access to company telephone directories and human resource manuals.

13.13 INTERNET SECURITY: A SPECIAL ISSUE

Not a cow, nor a gift of land, nor yet a gift of food, is so important as the gift of safety, which is declared to be the great gift among all gifts in this world.

Panchatantra, fifth century B.C.

The Internet is inherently insecure. It is, after all, an open network. Its openness certainly is one of its major strengths and, at the same, perhaps its major weakness. While e-commerce is promoted as the future of the Internet, this lack of security certainly has slowed development of commercial applications. It is worth noting, however, that providing your credit card number over the Internet is probably as secure as giving your credit card to a server at a restaurant.

Beyond security concerns over Internet commerce, you must remember that access to the Internet is a two-way door—just as users can get out, others can get in. All too often, those others have no legitimate right to do so. Although few organizations admit to having had their systems breached via the Internet, it is clear that such occurrences are all too common.

The Internet is rife with risks. Hackers, crackers, saboteurs, and other unsavory characters abound, eagerly attacking the Net and its users at every opportunity. The risks certainly include system intrusion, unauthorized data access, system sabotage, planting of viruses, theft of data, theft of credit card numbers, and theft of passwords. While the Internet and the Web cannot be blamed for the concepts and practice of mischief, fraud, theft, and other socially unacceptable forms of behavior, they certainly provide another high-tech cyberalley on the information superhighway.

The Computer Emergency Response Team (CERT) at Carnegie-Mellon University comprises a group of experts which are responsible for overseeing security issues on the Internet. While it is highly doubtful that a single security measure or standard will prevail in the near and distant future, there exist a number of options, including message encryption, authentication, and authorization. Firewalls, incorporating much of the above, recently have gained the spotlight in terms of a defense mechanism.

· Private-key encryption, also known as single-key or secret-key encryption, uses the same key for both encryption (encoding) and decryption (decoding). This approach requires that the key be kept secret through some form of secure key transmission prior to the ensuing data transfer.

· Public-key encryption involves the RSA encryption key that can be used by all authorized network users. The key for decryption is kept secret. Public-key encryption is much slower than private key, but the dissemination of the key is accomplished much more quickly. Public-key encryption is available freely on the Internet via a program known as Pretty Good Privacy (PGP), developed by Philip Zimmerman. PGP was under a cloud for some time because there was concern that it was so powerful as to violate U.S. technology export laws. By the way, encryption technology technically is classified under U.S. law as a form of munitions. The commercial version of PGP is known as ViaCrypt PGP, offering an improved user interface.

Data encryption standards include Data Private Facility (DPF), DES, RSA, and Clipper. Data Encryption Standard (DES), which uses a challenge-response approach and intelligent tokens, was formulated by the U.S. National Bureau of Standards. RSA, named after its developers, Rivest, Shamir, and Adleman, is for public key encryption. Clipper, an encryption standard developed by the U.S. government, uses escrowed keys to permit government deciphering through a back door. Clipper, which is nonexportable, is used extensively by the U.S. government and those who wish to do business with it. Encryption programs used on the Net include SSL, S-HTTP, and combinations of them.

· Secure Sockets Layer (SSL) from Netscape negotiates point-to-point security between client and server, including type of encryption scheme and exchange of encryption keys. SSL sends messages over a socket, which is a secure channel at the connection layer and existing in virtually every TCP/IP application. While SSL can accommodate a number of encryption algorithms, Netscape has licensed RSA Data Security's BSafe to provide end-to-end encryption as well as key creation and certification. Netscape's Netsite Commerce Server technology, including SSL, has been licensed by the likes of DEC (now part of Compaq, which is part of Hewlett-Packard), Novell, the Bank of America, and Delphi. Socket, by the way, is an operating system abstraction that permits application programs to access communications protocols automatically. Bolt Beranek and Newman (now BBN Technologies) developed this concept in conjunction with the company's early work on TCP/IP.

· Secure HyperText Transport Protocol (S-HTTP) from Enterprise Integration Technologies also negotiates point-to-point security between client and server, although at the application layer. EIT has licensed RSA Data Security's BSafe Toolkit for Interoperable Privacy-Enhanced Messaging (TIPEM). S-HTTP is a superset of HTTP and, therefore, is specific to the Web; several manufacturers of Web servers have announced plans to include S-HTTP in their products. S-HTTP has gained the support of the W3C and looks to be moving toward acceptance as a de facto standard.

· Password protection is imposed to restrict individuals on a site, host, application, screen, and field level. Passwords should be of reasonably long length, alphanumeric in nature, and changed periodically. There is a current trend toward the use of dedicated password servers for password management. Password Authentication Protocol (PAP) is a commonly used mechanism for password protection in support of remote users. While PAP is easy to use, passwords typically are sent to the Remote Access Server (RAS) in plain text (i.e., in the clear, or unencrypted).

· Intelligent tokens are hardware devices that generate one-time passwords to be verified by a secure server. They often work on a cumbersome challenge-response basis. Challenge Handshake Authentication Protocol (CHAP) is an example of this improved approach. CHAP involves the RASs challenging the remote user with a random number. The user responds with a digest, which is an encrypted password based on the random-number challenge. The RAS then decrypts the password using that same random-number key to verify the identity of the remote user.

Authorization provides a means of controlling which legitimate users have access to which resources. Authorization involves complex software that resides on every secured computer on the network; ideally, it provides single sign-on capability. Authorization systems commonly used in support of Internet security include Kerberos, Sesame, and Access Manager:

· Kerberos draws its name from Kerberos (also known as Cerebrus), the three-headed monster that guarded the entryway to the infernal regions in Greek mythology. Perhaps the best known authorization software, it was developed by the Massachusetts Institute of Technology (MIT) and is available free, although more powerful commercial versions exist as well. As Kerberos uses DES, it is not easily exportable. IBM's Kryptoknight is a weaker but exportable Kerberos variant. Although, according to Greek legend, Hercules defeated Kerberos, a hacker of Herculean proportions has yet to emerge victorious over this powerful software.

· Sesame (Secure European System for Applications in a Multivendor Environment) was developed by the ECMA (European Computer Manufacturers Association). It is flexible, open, and intended for large, heterogeneous network computing environments. It also is highly complex and not effective for smaller applications.

· Access Manager is an authorization mechanism approved by the IETF. Access Manager uses an API for application development, employing scripting. Scripting involves a process of mimicking the log-on procedures of a program, providing basic levels of security for small networks.

Authentication, as previously discussed, is a means of access control that ensures that users are who they claim to be. Whether through password protection or intelligent tokens, the authentication process is intended to avoid the possibility that unauthorized users might gain access to internal computing or network resources.

Tunneling is the process of encapsulating an encrypted data packet in an IP packet for secure transmission across the inherently insecure Internet, as illustrated in Figure 13.10. The four leading tunneling protocols are SOCKSv5, PPTP, L2TP, and IPsec:

· SOCKSv5 is an authentication technique that runs at the Session Layer (Layer 5) of the OSI Reference Model. Through the use of secure sockets negotiated between client and server over a virtual circuit and on a session-by-session basis, SOCKv5 supports the security of UDP datagrams as a data stream, rather than on a packet-by-packet basis. SOCKSv5 also supports protocol-specific communications, such as SMTP, to prevent hackers from extracting e-mail data through the use of an alias. SOCKSv5 is a cross-platform technique, working across multiple operating systems and browsers. SOCKSv5 also interoperates on top of IPv4, IPsec, PPTP, L2TP, and other lower level protocols.

· Point-to-Point Tunneling Protocol (PPTP) operates at the Data Link Layer (Layer 2) of the OSI Reference Model. Initially conceived by Ascend and developed by Microsoft and embedded in Windows NT, PPTP is a proprietary technique that encapsulates Point-to-Point Protocol (PPP) frames with IP packets. Packet filters provide access control end to end and server to server.

· IP Security (IPsec) as I noted earlier in this chapter, is the security mechanism developed for IPv6 and is used in IPv4 as well. In a dual-stack mode, IPv4 frames are encapsulated for transport within encrypted IPv6 frames. IPsec runs at the Network Layer (Layer 3). IPsec is a standards-based solution from the IETF and is defined in RFC 2401.

VPNs really are all about cost effectiveness, with the emphasis on cost rather than raw performance. Especially in the context of the IP-based VPN, the reduction in costs can be very significant indeed. Most such VPNs are based on the public Internet, although the IP-based networks from carriers such as Level 3 and Qwest also support VPNs. Access techniques include all of the options I have detailed, including dial-up modems, cable modems, ISDN, xDSL, Frame Relay, and ATM. The application scenarios include remote access, intranets, and extranets. Note that VPNs can be incredibly cost effective in support of the multinational enterprise, as truly private networks often are prohibitively expensive, if even available, on an international basis.

· Remote access VPNs are highly cost effective in support of telecommuters and mobile and remote workers. Assuming that the worker can reach the Internet or other IP-based network over a dial-up or other form of connection, reasonably secure communications can be accomplished with the home office or any branch office on the VPN. The level of bandwidth provided, of course, depends on the speed of the remote access link, the speed of the link to the home or branch office and the total volume of traffic over that link, and the level of congestion currently experienced over the shared IP backbone network. Significant cost savings can be realized when you compare VPNs to dial-up PSTN costs, whether based on Direct Distance Dialing (DDD) or toll-free access numbers. Each client PC, of course, must be equipped with software that supports the necessary tunneling protocols. Savings also can be realized at the branch, regional, or corporate location, as both T-carrier circuits connected to the PSTN and associated Remote Authentication Dial-In User Service (RADIUS) routers can be either eliminated or consolidated, in favor of shared T-carrier circuits connecting to the VPN.

· Intranet VPNs serve to link branch, regional, and corporate offices. Access to the VPN generally would be on the basis of dedicated circuits, usually in the form of unchannelized T-carrier, which might run the Frame Relay protocol or, perhaps, ATM. In this scenario, the elimination or consolidation of access circuits can yield significant cost benefits at all connected sites.

· Extranet VPNs serve to link vendors, customers, affiliates, and distributors into the main corporate office. An extranet VPN often must be extremely flexible in terms of access techniques and security mechanisms, as the corporate sponsor may have little control over the specifics of the external users'systems and processes.

13.14 MISUSE AND CONTENT

There have been a large number of highly publicized cases of the Internet and the Web being misused for illicit and immoral purposes. The Internet has been used for transmitting stolen credit card numbers and cellular telephone ID numbers, for example. While any communications medium can be used for such purposes, the Net creates another set of difficulties for law enforcement because communications are virtually instantaneous and multiple parties can gain access to the illegal data through a bulletin board. Further, it is difficult, if not impossible, in many cases to track down the offenders.

More significant from a social perspective is the fact that Internet chat rooms have been used to lure minors into the clutches of pedophiles and others who wish to take advantage of them. The offenders generally seem to mask themselves as minors, striking up an electronic conversation and suggesting a meeting somewhere. While there have been only a relative few such cases uncovered and publicized, the risks exist and the consequences can be terrible.

Also of great concern is the issue of content and access to it. Numerous websites contain sexually explicit material, including photographs. While adults have a constitutional right to view such material, minors generally do not; the Internet and the Web really have no effective means of controlling access, since the users are anonymous and their ages and other characteristics are unknown. Most of the truly explicit websites (so the author is told) offer very little for free, with full access to the offending material generally being provided only on the basis of subscription paid by credit card.

Outside the United States, very tight content controls have been exercised in some countries. As you might expect, those nations include Singapore and the Peoples Republic of China (PRC). In January 1999, an Internet e-mail broker in the PRC was sentenced to two years in prison for inciting the overthrow of state power by supplying 30,000 e-mail addresses to a prodemocracy magazine in the United States. The sentence was considered by many to be light, as three leaders of the China Democracy Party received sentences of 11, 12, and 13 years after being found guilty of the same charge in December 1998. Since that time, there have been a number of similar occurrences in various countries.

13.15 INTERNET ODDITIES, SCREWBALL APPLICATIONS, AND SOME REALLY GOOD IDEAS

As mentioned previously, only technology and the human mind can limit the applications for the Internet and the Web. There have been a number of recent announcements of Internet oddities and screwball applications as well as some really good ideas. I've got my own ideas about which are which, and I'm sure you have yours.

· Package tracking is offered on the Web by FedEx, UPS, USPS, and other carriers. Customers can access the websites and get the latest status of packages, based on package tracking numbers. The savings easily are in the millions of dollars, compared to the costs of handling voice calls through human agents or even interactive voice response applications in an incoming call center.

· Document delivery over the Internet has been promoted by UPS, through its document exchange service known as UPS OnLine Courier. Senders upload documents to the secure UPS server, at which point UPS sends e-mail notifications to the recipients, advising them of documents awaiting delivery. The recipient uses the URL provided in the notification to download the document from the secure server via a Web browser. An option enables the sender to require that the recipient use a password known only to the sender and recipient. Security is provided via 128-bit encryption on the server and 40-bit Secure Socket Layer (SSL) encryption during transport. Pricing is attractive, especially when compared to U.S. Postal Service (USPS) Registered mail and traditional courier services. Although UPS discontinued OnLine Courier effective September 1, 2002, it is a good idea that surely will find a market at some point in the future. Over the past few years, the USPS has offered a number of e-mail services, some of which have experienced modest levels of success and some of which have been discontinued.

· Distribution of software, upgrades, and bug fixes is a really good idea. In many cases, software is provided free of charge as a public domain release, with subsequent, enhanced versions provided commercially.

· Cyberfairs are job fairs that enable companies to post job listings and recruit applicants online. Interactive communications are supported.

· Cybergambling (Now there's a constructive idea!) is now offered by casinos. The casinos can circumvent U.S. gambling laws by placing their servers offshore. Users are required to have offshore bank accounts, as well, to deposit their winnings. imho (that's netspeak for in my humble opinion. note the lack of capitalization. that's also an e-mail convention. it's in the style of e.e. cum-mings.), we do not need a Virtual Vegas. The U.S. government appears to agree. Actually, the State of Washington outlawed cybergambling in 2006.

· Cybersex is something I mention only because its popularity demands that I do so. It is not worthy of further comment, at least not in my humble opinion.

· Music distribution over the Internet currently makes use of MP3 (MPEG-1, Audio Layer 3) compression algorithm. As each minute of uncompressed digital music requires a file of roughly 10 MB, it takes a very long time to download an entire CD-ROM album, even over an ADSL connection. MP3 compresses a 600-MB CD-ROM music album down to about 50 MB. Issues of copyright infringement are significant with respect to this application.

· Online auction houses offer electronic auctions for just about everything imaginable, from airline tickets, hotel rooms, and cruises to sporting goods, antique china, and fabric remnants for quilting. The risks are several: You do not get to see the goods before you buy, and you have no real assurance of delivery. In other words, you buy solely on faith. Personally, I don't get it. I generally can get much better fares, much better flight schedules, and much better service from my travel agent. As it seems that the airlines are doing everything they can to put the travel agents out of business, I may not have that option forever. Margaret, my lovely wife, on the other hand, loves this sort of thing. Apparently, a lot of other people love this sort of thing as well. Beware, however. The U.S. Internet Crime Complaint Center (IC3) received 231,493 complaints in 2005. Internet auction fraud was by far the most reported offense, comprising 62.7 percent of referred complaints [61].

· Online grocers attracted a lot of attention several years ago. WebVan and HomeGrocer, the most prominent members of this group, offered the convenience of shopping from your PC over the Internet. In other words, you could shop from home or office for exorbitantly overpriced groceries that would be delivered right to your doorstep. Believe it or not, I actually know people who used these services, and they were very disappointed when both WebVan and HomeGrocer went bankrupt. Frankly, it did not surprise me a bit, although it does surprise me that at least one online grocer has survived. Peapod delivers to the Chicago, Illinois, Milwaukee, Minnesota, and southeast Wisconsin areas. Through an alliance or strategic relationship of some sort with Stop & Shop, it also supplies portions of Connecticut, Massachusetts, Rhode Island, New Jersey, and New York. Through the Giant grocery chain, it also offers service in the greater Washington, D.C., area. Is it any wonder that the latest government studies show that 62.5 percent of Americans are overweight or obese?

Having said all that, I also have to tell you that the first edition of this book, along with a number of other books, was once published on the Internet by a university in a nation that once was one of the Soviet republics. It was an obvious case of copyright infringement that took a good deal of time, effort, and expense to stop. It was outright thievery that I did not appreciate a bit, and neither would you, were you in my position.

You may recall the horror of the TWA Flight 800 incident. You also may recall the statements of Pierre Salinger, who was the press secretary for President John F. Kennedy. Salinger claimed to know on national news that Flight 800 was, in fact, shot down by an errant missile fired from a U.S. Navy warship. Salinger got this bogus information from a website.

You may remember the Good Times virus and a host of other viruses. While there are many real viruses that pose real threats, many of them are absolute hoaxes, the rumors of which spread like wildfire over the Internet. Virtual panic resulted.

You may remember that in June 1998 an Associated Press reporter mistakenly posted a prepared obituary for Bob Hope on that wire service's website. While the error was discovered virtually immediately and the obituary was removed within 15 min or so, a U.S. congressman saw it and eulogized Bob Hope from the floor of the House of Representatives. Mr. Hope's response was somewhat along the lines of Mark Twain's, "Reports of my death have been greatly exaggerated." Perhaps Sir Walter Scott said it best: "Oh, what a tangled web we weave when first we practice to deceive."

The bottom line is that it is your responsibility to seek the true and the beautiful. There are no filters on the Internet or the Web that truly will protect you. Protect yourself if you can, and please protect your children, for they often cannot—or will not—protect themselves.

Chapter 14: Network Convergence

Arthur C. Clarke, United Nations Telecommunications Day, 1983

While Arthur C. Clarke may have overstated things a bit in 1983, his statement seems to be pretty accurate in 2006. Virtually anything we want to do is possible, short of teleportation, and scientists are working on that, or so I'm told. The limitation may well be that we just do not want to do enough. In other words, our imaginations are probably more limited that our abilities to satisfy them through the invention of new technologies. Ultimately, of course, technology really is just an enabler of applications, and people keep dreaming up new and exciting applications, many of which are highly demanding technically. Many of these applications are extremely bandwidth intensive, including high-speed data, video, and even multimedia. Billions of dollars have been spent on the development and deployment of new network infrastructure in an attempt to satisfy our seemingly insatiable desires to communicate instantly and in a variety of formats, from audio to data to video to multimedia.

Analog has given way to digital technology. Copper has yielded to glass in the backbone and is finding its way into the local loop. Wireline networks have given way to wireless, at least in support of mobile communications, and wireless technologies do a wonderful job of supplementing the wireline backbone networks. Wireless Local Area Networks (WLANs) have been standardized and are enjoying great popularity. Circuit switching is challenged by packet switching across all applications types. Satellite constellations support communications anywhere on the face of the earth. The Internet and the Web provide access to virtually any type of data in any database residing on any networked computer anywhere in the world—issues of security not withstanding and assuming that economics, laws, politics, culture, and religion do not get in the way. Electronic commerce has changed the way we shop for everything from books to music to clothing to automobiles to groceries. Telephone calls are so inexpensive that we no longer give any thought to picking up the phone and calling across the country and even across continents. Telephone calls, snail mail, and faxes have largely yielded to electronic mail, which is virtually free, assuming Internet access. Plain old telephone sets are yielding to softphones and cell phones, and the size of the cell phones we use seems to be related inversely to the size of the SUVs (Sports Utility Vehicles) we drive. By the time you buy your next Chevy Subdivision, your cell phone may well fit in your ear and be activated by your brain waves, so be careful what you think. Note: That cell phone likely will cost less than a tank of gas. (Actually, my cell phone was free, although it doesn't quite fit in my ear.)

The U.S. economy has changed in the last 200 years from agrarian to industrial to informational. Ten years ago, we lamented the fact that we no longer makeanything in this country and that our having lost our industrial edge would be the ruination of the American way of life. In fact, generation X could look forward to being the first generation in our history to enjoy a lower standard of living than the previous generation. While we no longer make every part that goes into every car that we assemble, buy, and drive, we do create much of the information technology that drives the rest of the world that makes the parts that go into the Chevy Subdivisions and other vehicles that are assembled in Mexico or Canada or who knows where and shipped to the United States and bought over the Web that runs on technology that was developed mostly in the United States and is embedded in machines that we were at the forefront of inventing and which include components we invented and bought over the Web; and the same goes for the cell phones that soon will fit in our ears so that we are not distracted while driving our Subdivisions—and on and on and on.

Although the term information superhighway has fallen out of favor and is even considered quaint, the concept remains sound. The applications are exciting and even compelling. Many of the enabling technologies have been invented, many standards are in place, and billions of dollars are being invested in infrastructure. Technology is moving forward and propelling us along with it at ever-increasing speed. Progress marches on. Resistance is futile. At this point, it is worth reflecting on the content of a letter reportedly sent in 1829 from Martin Van Buren, Governor of New York, to President Andrew Jackson:

Dear Mr. President:

The canal system of this country is being threatened by the spread of a new form of transportation known as railroads. The federal government must preserve the canals for the following reasons.

One: If canal boats are supplanted by railroads, serious unemployment will result. Two: Boat builders would suffer and tow-line, whip and harness makers would be left destitute. Three: Canal boats are absolutely essential to the defense of the United States.

As you may well know, Mr. President, railroad carriages are pulled at the enormous speed of 15 miles per hour by engines which, in addition to endangering life and limb of passengers, roar and snort their way through the countryside, setting fire to crops, scaring the livestock, and frightening women and children. The Almighty certainly never intended that people should travel at such breakneck speed.

(signed)

Martin Van Buren

Governor of New York

14.1 CONVERGENCE DEFINED

Convergence is defined as the moving toward union or one another. In the context of the information superhighway, revolution, evolution, or whatever you choose to call it, the concept of convergence cuts across a number of dimensions, including a wide range of applications and the underlying technologies. In full form, convergence represents the coming together of every technology and application discussed in the previous 13 chapters.

Applications, truly, are at the very crux of convergence. The only conceivable business reason for investing billions of dollars in network technologies is to serve revenue-producing, profit-generating applications. (There are, of course, a number of social, cultural, and political reasons for either encouraging or discouraging such investments.) There appears to be no single killer app driving the information superhighway—although the Web comes close, it is really a means of serving a vast collection of applications. Certainly, a number of interesting and productive niche applications exist which, in various user-specific combinations, constitute a killer app suite. One of the most compelling applications is that of Web-enabled call centers, as I discussed previously. Multimedia, in full form, currently is viewed as the ultimate in terms of presentation mode, although compelling applications of real substance have yet to be defined.

Data networks are digital in the backbone and have been so for years. While residential and small-business users typically access the network on a dial-up basis through modems, larger users typically make use of digital local loops in the form of T-carrier. The generic Digital Subscriber Line (xDSL) local loop technology is making a huge impact, although availability remains something of an issue. Frame Relay (FR) is on the decline, no major new investments are being made in ATM, and the undeniable shift to IP is taking place quickly.

Community Antenna TeleVision (CATV) networks also are experiencing significant upgrades. The old one-way analog coaxial cable distribution systems are being converted to two-way digital and the coaxial distribution cable systems are being replaced with fiber. The CATV networks also are easing out legacy protocols in favor of I P. Additionally, some of the satellite TV networks have been upgraded to support two-way Internet access, and IP is the protocol of choice.

Cellular networks have almost completely shifted from analog to digital. Speeds have increased considerably with the introduction of 2.5G and 3G technologies that support not only voice but also data and even video. Cellular networks are one of the few domains in which the trend is not toward IP, although IP gateways certainly support the necessary internetwork interfaces.

Local Area Network (LAN) technology has made incredible strides in the past decade or so. Ethernet overwhelmed Token Ring, Fiber-Distributed Data Interface (FDDI), and other protocols, switches replaced bridges and hubs, and speeds increased from 10 Mbps to 100 Mbps, 1 Gbps, and even 10 Gbps. Priority mechanisms developed to support Grade of Service (GoS), if not guaranteed Quality of Service (QoS). In combination, the speed and GoS mechanisms have made switched Ethernet entirely appropriate for voice as well as data. In fact, switched Ethernet LANs have become the platform for integrated voice/data Private Branch eXchanges (PBXs), which, by the way, run the Internet Protocol.

WLANs became commonplace with the standardization of wireless Ethernet as 802.11a/b/g, and WLAN speeds now reach 54 Mbps, at least theoretically. 802.11n is soon to make an appearance, doubling theoretical speeds to 108 Mbps. Wi-Fi WLANs have now replaced wired LANs in many cases, and Voice over Wi-Fi (VoWiFi) is beginning to make serious sense.

14.2 DRIVING FORCES

The evolution of the network moved at a relatively glacial pace for the first hundred years or so. With the exception of Step-by-Step (S × S) electromechanical switches and rotary dial telephones, little in the way of technology was introduced for the first 50 years. The introduction of Crossbar (Xbar) switching in 1937, Direct Distance Dialing (DDD) in the 1950s, and tone dialing in the 1960s were considered absolutely revolutionary. The pace accelerated with DDS, Switched 56 kbps, and T-carrier services in the 1960s, 1970s, and 1980s. Toward the mid − 1980s, the pace picked up considerably, and ISDN deployment began to make an impact in the early 1990s. In the last 10 years, the pace picked up even more and became a full-tilt race through April 2000. At that point absolutely awful fiscal policy in the United States, combined with irrational exuberance in the technology sector of the stock markets, general weakness in the worldwide economy, and a number of other factors created a worldwide recession. Things got back on track in 2004 or so, although the business of telecommunications changed a good deal in the meantime. Forces driving the development of the convergence scenario include deregulation, privatization and competition, applications, and technology.

Deregulation, perhaps, was the primary driving force. Beginning with the Federal Communications Commission (FCC) Carterfone decision in 1968, end users were presented with a wide variety of options for terminal equipment. The Modified Final Judgment (MFJ), which took full effect on January 1, 1984, was the next step, dictating the breakup of the AT&T Bell System and ending what had constituted a virtual monopoly over communications in the United States—from research and development, to network equipment manufacturing, to service delivery. Competition developed as a result of the actions of the FCC and the federal courts. That level of competition has intensified by orders of magnitude since the passage of the Telecommunications Act of 1996 and now includes competition for local service and broadband Internet access as well as equipment and long-distance services.

Customer Premises Equipment (CPE) and long-distance competition developed quickly after the Carterfone decision and the MFJ. CPE competition is widespread, with literally thousands of manufacturers competing for the voice, data, video, and image systems markets. Competition in the long-distance business is intense in the United States. In 2002 there were well over 400 facilities-based carriers vying for the interLATA (Local Access and Transport Area) market, although most of them have since gone out of business or been acquired by the incumbent carriers. The ILECs also acquired IntereXchange Carriers (IXCs) and now compete against each other intensely.

The cost of an integrated terminal device certainly can be less than the cost of multiple devices. Assuming that you have a requirement for a tele phone set, a computer, a videoconferencing unit, and a television set, the cost of a single, multifunctional device can be less than the total cost of the individual devices. The cost of a single, multifunctional local loop can be less than the cost of multiple, application-specific loops. And the cost of a single, multifunctional network can be less than the cost of multiple, application-specific networks. A basic assumption is that the technology is in place to enable such an integrated scenario—and that enough people are sufficiently interested to buy enough integrated terminals, rent enough loops, and subscribe to enough network services to enable the manufacturers to build and sell enough devices and develop and sell enough application software to bring the unit cost down to an affordable level. Only then can the network providers offer such services at affordable cost.

The public Internet clearly is being positioned as the converged network, at least for the great unwashed masses of us who cannot afford or justify private IP networks. The volume of Internet traffic has grown to incredible proportions and increasingly is burdened with ever more bandwidth-intensive applications such as graphics and image transmission as well as QoS-sensitive applications such as voice and video. The end result is an Internet that is increasingly stressed at all levels. So, the backbone service providers and ISPs alike are expected to add more capacity in order to meet end-user expectations, but it is not clear how they will recover their costs. One proposal is for a tiered, content-sensitive, and discriminatory pricing structure that charges end users more for QoS in support of applications such as VoIP, streaming video, Virtual Private Networks (VPNs), and even the Web, in general. This has sparked considerable controversy, particularly in the United States, where the Internet has always been a public and content-neutral medium. The controversy over network neutrality is unlikely to be resolved in the near future.

Voice communications, including both the provisioning of local loops and the delivery of local service, certainly not only is part and parcel of convergence but also represents its very foundation. While local service is not stunningly profitable, the successful local service provider has a competitive edge with the consumer with respect to long-distance service. Bundled with other voice services such as enhanced custom calling features and broadband Internet access, such a package can be most attractive to the user and highly profitable to the service provider, especially when delivered over a single Asymmetric Digital Subscriber Line (ADSL) or perhaps PON local loop.

Data communications services tend to be highly profitable, even in these highly competitive times. The historical growth of—and growth potential for—bandwidth-intensive data services is well documented. In the competitive market for data communications services, emphasis is on highly profitable dedicated services such as native LAN-to-LAN connectivity as well as on enhanced and unregulated services such as Frame Relay and IP-based VPNs.

· Airline, hotel, and rental car reservations

· Games

· Gambling

Currently, CATV providers offer television on a highlyprofitable basis, especially since rate regulations were lifted in recent years. In fact, CATV providers essentially are unregulated at the federal level and only marginally regulated at the local level. In addition to basic offerings, premium channel subscription, Pay-Per-View (PPV), and Video-On-Demand (VOD) are extremely popular and profitable. As I discuss in previous chapters, CATV networks rapidly are being upgraded in support of Internet access and voice communications. Videocon ferencing, multimedia, and other applications also can be supported over CATV networks.

While CATV is likely to continue as the primary means of TV delivery in the United States, Direct Broadcast Satellite (DBS) has grown in popularity, offering greater choice of channels at reasonable cost. Employing digital satellites and MPEG − 2 compression, transmission quality is excellent, although propagation delay remains an issue with traditional Geosynchronous Earth-Orbiting (GEO) satellites.

Telco PON networks currently in the early stages of deployment in the United States are capable of supporting entertainment TV as well as voice and broadband Internet access. While the telcos have yet to offer TV programming over those networks to any appreciable commercial extent, that certainly is the intent in the longer term under the triple-play concept of converged voice, Internet access, and entertainment TV. Verizon is already promoting entertainment over its FiOS PON offering.

The performance of an integrated suite of applications certainly is an improvement over that of multiple, disconnected applications. This is the essence of multimedia. We do not necessarily need to view text, or image, or video information in connection with every telephone call, but it often is advantageous to do so. When you talk to a realtor in another city about an impending house-hunting trip, it is helpful to see images of houses that meet your criteria, see a map of the city to get a sense of the house's location, view the proposed contract with the realtor, and even see the realtor through a videoconference. You certainly can do most of these things through a combination of a telephone call, a fax transmission, and an e-mail transfer, but combining them all together in a single, interactive, multimedia presentation enhances the overall performance of the communication.

Videoconferencing expands to video dial tone, or visual dial tone, in a full-on broadband convergence scenario. This application appeals greatly to some and borders on the ridiculous to others. Affordable bandwidth and terminal equipment remain major issues. Once the cost becomes reasonable and the CPE configuration issues are resolved, videoconferencing will become commonplace.

· PSTN: The conventional circuit-switched PSTN offers unyielding voice quality and will be in place for many years to come. Ubiquitous access to this legacy public voice network is essential for the foreseeable future.

· Internet: The public Internet will continue to be the network by which we access the Web and by which we send public e-mail. It also increasingly will serve as a low-cost alternative for voice using VoIP.

· Private IP Network: Carriers will offer access to well-designed and carefully managed private IP networks for business-class voice, VPN service, and other business-class or premium services that benefit from increased bandwidth, improved congestion management, and enhanced security.

14.4 THE RACE IS ON: MERGERS AND ACQUISITIONS (M&As)

The race is on and here comes pride up the backstretch. Heartaches are goin' to the inside. My tears are holdin' back. They're tryin' not to fall. … And the winner loses all.

The Race Is On, composed by Don Rollins and first recorded by George Jones in 1965

ILECs, CLECs, IXCs, CAPs/AAVs, CATV providers, PCS licensees, satellite service providers, electric utilities, and others all have jockeyed for the pole position to lay the grid and provide the services that will make them and their stockholders wealthy. In recognition of the fact that convergence is the name of the game and that no one company has all of the answers, hardware manufacturers, software developers, and carriers all have been gobbling each other up at a record pace. The race definitely is on! The following is a small but representative sample of the scope and scale of such activity. I find it relevant in the context of convergence for several reasons. First, there has been a lot of consolidation (i.e., convergence) in the corporate world as companies have merged and acquired each other. Within the domain of the original AT&T Bell Systems, there has been a great deal of fragmentation (i.e., divergence) and reassembly (convergence). All of that activity has led to increased levels of competition at times and decreased levels at other times. One way or another, this activity has defined, at least in part, the nature and form of convergence.

Break up to make up, that's all we do. First you love me, then you hate me. It's a game for fools.

Break Up To Make Up, The Stylistics, 1973

Bell Atlantic got busy, merging with NYNEX in August 1997. NYNEX ceased to exist. In July 1999, Bell Atlantic acquired GTE, and renamed the company Verizon.

US West reached a definitive agreement in 1999 to merge with Global Crossing, an upstart international submarine fiber-optic carrier. Global Crossing also reached a definitive agreement to merge with Frontier, a LEC and IXC that began life as Rochester Telephone. Qwest, an upstart IP-based IXC, then made a hostile bid for both US West and Frontier and a war broke out. Actually, it was more of a skirmish, as Qwest very quickly (July 1999) was declared the winner with respect to the US West merger. Global Crossing walked away with the Frontier merger agreement remaining intact. Global Crossing and Qwest were attracted to US West because a good deal of long-distance traffic either originates or terminates in US West territory. While Frontier's properties in Rochester, New York, were not particularly attractive, its status as a second-tier IXC was very attractive. The bottom line is that both Global Crossing and Qwest spent billions of dollars laying pipe (i.e., building backbone infrastructure), which they needed to fill with minutes of traffic. It is faster and easier and often less expensive to buy the minutes than it is to take them away from the competition. Qwest has a history of this sort of activity, having merged with LCI in June 1998; LCI previously (September 1997) acquired USLD Communications. US West previously spun off its CATV business, US West Media Group, which became MediaOne, which was subsequently merged into AT&T. By the way, Global Crossing sold the ILEC portion of Frontier to Citizens Communications in July 2000.

Who owns AT&T? The stock of American Telephone & Telegraph Company, parent company of the Bell System, is held by more than 750,000 shareholders. These stockholders are about equal to the population of the city of Pittsburgh or the state of Rhode Island.

Telephone Almanac, Bell Telephone System, 1949

On January 1, 1997, AT&T effected the largest voluntary breakup in history. The US $ 75 billion company split into three market-focused companies, also selling AT&T Capital, its captive financing business. Approximately 8500 employees, all in the Global Information Solutions (GSI) computer business, lost their jobs fairly immediately. GSI resulted from the NCR acquisition, which did not live up to expectations. Hundreds of thousands of others lost their jobs over time. The post-divestiture AT&T boasted assets of US $ 79.2 billion, annual revenues of US $ 75.1 billion, and a total workforce of 303,000, which was down from over 1,000,000 prior to divestiture. Table 14.2 provides a view of AT&T, both postdivestiture and immediately postspinoff.

After the voluntary breakup, AT&T kept busy, of course. In February 1996, AT&T announced its Internet access service, which was an instant hit. AT&T World-Net service suffered considerably when faced with competition from the likes of AOL, MSN, United Online, and Prodigy and never recovered. The brand still exists and is marketed outside the AT&T (nee SBC) franchise service areas but is not in the top 22 Internet Service Providers (ISPs) in the United States, according to ISP Planet.

In January 1996, AT&T purchased a 2.5 percent stake (US $ 137 million) in DirecTV, a provider of entertainment TV via DBS satellite. AT&T marketed DirectTV to its customer base, billing for the monthly services through its standard long-distance billing system. Purchase and installation of the dishes also could be financed through AT&T credit cards. Subsequently, AT&T divested its stake in Direct TV.

In July 1998, AT&T and British Telecom (BT) announced the formation of Global Venture, an international alliance that expected revenues of US $ 10 billion in its first year of operation. BT contributed to Global Venture its Concert services, which it previously linked with MCI. BT, at one time, owned 20 percent of MCI, which subsequently merged with Worldcom to become MCI Worldcom. AT&T's link with BT required that it dissolve its WorldPartners WorldSource international alliances, which involved a large number of partners all over the world. Global Venture subsequently was renamed Concert. On October 16, 2001, AT&T and BT announced their decision to dismantle Concert but to honor existing contracts and service-level agreements for three years.

In 2006, the tattered remnants of AT&T were acquired by SBC for approximately US $ 16 billion, which named the combined entity AT&T. In just over 20 years, one of the oldest, largest, and most respected companies in the world was reduced to a ghost, at least partially due to the sheer incompetence of some members of its senior management. On a personal note, I thank my lucky stars that I left the Bell System of my own free will long, long before AT&T collapsed. Heck, I never did fit in, anyway. I reckon I didn't have enough of aBell-shaped head, as we used to say.

14.5 ONE POTATO, TWO POTATO, THREE POTATOE, FOUR …

Potato(e)can be spelled more than one way, as then-Vice President Dan Quayle illustrated several years ago. There also is more than one way to build an information superhighway. There is a very serious question as to just how many wires(hardwires and wireless wires)should be extended to provide access to the networks in a convergence scenario. After all, the very word convergence means coming together. You have to wonder how much redundant infrastructure is reasonable and at what cost. Perhaps the electric utilities have the right idea—build one information grid and provide access to a wide variety of service providers. Of course, this idea is not new. Utilities traditionally have provided a single grid for telecommunications, data communications, CATV, electric power, gas, water, and sewer services. The fundamental economic concept of a natural monopoly served by a common carrier still has merit.

It now seems as though a natural duopoly or triopoly will prevail in many areas at the local loop level. Telecommunications and CATV local loop networks already exist and have been upgraded in many areas. While the level of capital investment is not trivial in either case, it is quite likely that two grid providers can survive quite comfortably and profitably. If you add a wireless carrier or two to the equation through Wi-Fi or WiMAX technology, perhaps that number expands to three or four, given the flexibility of wireless network configuration. Broadband over Power Line (BPL) conceivably adds another wireline carrier to the mix, although BPL likely will be reserved for remote rural applications. But it is highly unlikely in the foreseeable future that there will be more than two or three wireline grids and perhaps a wireless bypass alternative except in small geographic areas where there exists a high concentration of substantial business users.

That is all at the Physical Layer (Layer 1), of course, but that is a grid issue. The service providers using the grid is another issue altogether.

· CATV Loops: The CATV networks will likely be closed to competition—they always have been and there is no reason to expect that will change in the foreseeable future. It also is highly unlikely that another CATV provider will build a competing wireline network. The telcos, however, will increasingly compete head on with triple-play services over PON.

· Telco Loops: The telco local networks were closed for 120 years, until the Telecom Act of 1996 opened them to competition. Now they seem to be closed again, for all practical purposes, with the recent designation of DSL as an information service. The RBOCs continue to wholesale DSL loops to competing ISPs, but that will not necessarily continue forever. PON local loops are similarly closed and will remain so. Full-on PON is an overlay network, which creates an interesting situation. Verizon is building overlay PON networks in New York City and a number of other major cities in the northeast United States. Those PON loops, of course, will be much more capable, much less maintenance intensive, and much more profitable that the UTP loops right alongside them. The PON loops likely will have unregulated status at the state as well federal level while the UTP loops will remain regulated, except for DSL, which Verizon will transition to PON. The PON loops likely will be maintained by the nonunion, or quasi-union, contractors who built them, while the regulated UTP voice loops will continue to be maintained by the solidly embedded members of the Communications Workers of America (CWA) and the International Brotherhood of Electrical Workers (IBEW). So, I am willing to bet the royalties from this book that Verizon will sell off the legacy regulated local exchange business, including the copper local loop, and will retain the nonregu-lated PON loops, over which it will deliver the triple play of IP voice, Internet access, and IPTV.

· Access Broadband over Power Line: Access BPL largely will remain a niche solution for remote rural situations. Access BPL will remain targeted primarily at broadband Internet access and voice.

While some of us remember certain aspects of the Bell System days of yore with a certain fondness, I think we all agree that competition has been a good thing for the most part. Competition may have gotten out of hand at some point, but the impacts largely were positive. The failures of the (old) AT&T, MCI, and Sprint as CLECs were undoubtedly their own fault, but it is a shame, for their losses are our own as well. Three once powerful competitors effectively disappeared.

14.6 NEXGEN CONVERGENCE: WIRELINE AND WIRELESS NETWORKS

Next-Generation (NexGen) convergence goes beyond the local loop alternatives of CATV, telco UTP and PON, WLL, and Access BPL. NexGen convergence also goes beyond the core network alternatives of the circuit-switched PSTN, the Internet, and private IP networks. NexGen convergence folds 2.5 G and 3 G cellular networks into the mix. Such a converged network is more than a gleam in some engineer's eye or a spark in some marketer's mind; it is taking form through the IP Multimedia Subsystem (IMS) architectural concept.

IMS originated in the 3rd-Generation Partnership Project (3GPP), which was seeking a common means by which GSM cellular operators could deliver data services. IMS subsequently transcended the cellular domain and is now being embraced by both wireless and wireline service providers. Industry groups such as the Multiservice Switching Forum (MSF), European Telecommunications Standards Institute (ETSI), and Alliance for Telecommunications Industry Solutions (ATIS) have adopted IMS as the foundation for their next-generation infrastructure strategies.

In the pre-IMS world, the PSTN and the Internet are very different, as we have discussed. The circuit-switched PSTN conforms to an architecture that centralizes the intelligence in the network. The Advanced Intelligent Network (AIN) places each service in an intelligent node and shares applications and communicates instructions to less intelligent switching platforms through Signaling System 7 (SS7). PSTN terminal devices are dumb. The packet-switched Internet distributes intelligence among clients and servers in support of peer-to-peer communications. Internet signaling and control are through the TCP/IP stack, with movement toward the Session Initiation Protocol (SIP) for applications such as VoIP. The cellular networks generally are circuit switched in nature and conform to the PSTN model of centralized intelligence and dumb terminals, with SS7 signaling and control.

An IMS-capable multimedia client connects over an IP-based core network to application and content servers that provide various value-added services such as multimedia conferencing and unified messaging. The client can operate over a PSTN connection, a wired Ethernet port, a Wi-Fi channel, a WiMAX channel, or a cellular channel and can roam between any of them. The IMS network will recognize the changing nature of the network and adjust accordingly, perhaps enabling video when the user connects to an 802.11g Wi-Fi network that will support the bandwidth requirement and disabling it when the user gets out of range of the Wi-Fi network but maintains connectivity through a handoff to a GSM cellular network that will not support the video bandwidth requirement. IMS manages internetwork handoffs, bandwidth negotiation and QoS, while it keeps peers engaged in the session advised via SIP as to the specifics of the level of multimedia presence [2 − 6].

So, that all translates into one device with one address that can be used just about anywhere at just about any time under just about any circumstances and that communicates with its peers through a network that can adapt to just about anything. That seems like convergence to me.