Case Study 3: Design Scenario - C

Example of a WAN Design

This section presents a WAN design that was developed using some of the design steps in this book. The section describes an actual network design that was developed for Klamath Paper Products. The name of the company has been changed. The example is based on a real network design, but some of the facts have been changed to preserve the privacy of the company and protect the security of the company’s network, and to make it possible to present a simple and easy-to-understand example.

Background Information for the WAN Design Project

Klamath Paper Products, Inc. manufactures paper and packaging products, including office paper, newsprint, cartons, and corrugated boxes. They also manufacture wood pulp and chemicals used in the manufacturing of pulp and paper.

Klamath Paper Products (which will be called Klamath from now on) has approximately 15 sites in the western United States. Headquarters are in Portland, Oregon. Klamath employs around 1500 people and has customers all over the world, with a large customer base in Asia.

Klamath is concerned about reduced profit margins caused by fewer sales in Asia in recent years and the scarcity of lumber used to manufacture Klamath’s products. Klamath recently completed a strategic re-engineering project that identified ways to increase profits by improving the efficiency of internal processes and making more use of recycled postconsumer paper in the production of new paper products.

As a result of the re-engineering project, the Conservation Initiative Task Force at Klamath plans to roll out an ongoing distance-learning program that will train all employees on ways to conserve raw materials, use recycled materials, and work more efficiently. Executive management considers the new training program vital to the continued success of Klamath, and approved funding to equip the training rooms at most sites with digital videoconferencing systems.

After Klamath installs the videoconferencing system and the WAN to support it, there are plans to offer classes to other companies in the wood and paper manufacturing industries. Klamath has recognized a business opportunity associated with the federal government’s plan to help pay for workers in the timber industry to attend classes in modern methods for sustainable forest management and environmentally sound lumber and paper production.

Business and Technical Goals

Klamath’s main business goals for the WAN design project are as follows:

· Increase profits by implementing a WAN that will support the goals of the Conservation Initiative Task Force, in particular the new distance-learning program.

· Improve the performance of the existing WAN to support more efficient operations.

· Contain the rising costs associated with operating the existing WAN.

· Provide a network that will let employees more easily share ideas for further improving efficiency and increasing the use of recycled materials.

· Provide a new source of revenue from the timber-industry distance-learning program.

Engineers in the telecommunications and networking departments added the following technical goals:

· Update the capacity and QoS capabilities of the existing WAN, which in its current state cannot support the new videoconferencing system.

· Design a network that uses currently available technologies from the WAN service providers in the region.

· Provide a network that offers a response time of 1/10th of a second or less for interactive applications.

· Provide a network that is available 99.98 percent of the time and offers an MTBF of 4000 hours (about 5.5 months) and an MTTR of 1 hour (with a low standard deviation from these average numbers).

· Improve the manageability of the network by simplifying the topology, which is currently a complex mesh of voice and data circuits.

· Design a network that will scale as new high-bandwidth applications are added in the future.

· Design a network that can support voice traffic in the future.

Network Applications

The new distance-learning application will use a two-way compressed digital video service based on the H.323 standards for videoconferencing. Each site with a training room will be equipped with a high-end digital video camera. Both synchronous and asynchronous distance learning will be supported. With synchronous distance learning, remote students attend classes taught by instructors at headquarters or other sites in “real time.” With asynchronous distance learning, students can check out a video class from a video server at headquarters and have the video transmitted to their site.

Other applications in use at Klamath include the following:

· The manufacturing support system runs on a mainframe in Portland. The system keeps track of manufacturing schedules and work orders. Members of the various manufacturing departments access the system from their PCs. This system is considered critical to Klamath’s mission to deliver products by the dates that were promised to customers.

· The financial modeling system runs on UNIX and makes use of an Oracle database that resides on UNIX servers in Portland. Financial analysts use applications on their PCs to access this system.

· The sales order-entry and tracking system runs on Windows servers. Sales and marketing personnel use their PCs to access this system.

· The graphics production system runs on Macintosh computers and uses Apple Filing Protocol (AFP) servers.

· Most users also deploy a standard set of desktop applications that includes email, calendaring, web browsing, file sharing, and printing. These applications use TCP/IP and NetBIOS.

User Communities

Table 11-4 shows a summarized view of the user communities at Klamath.

Table 11-4 Klamath User Communities

User Community Name	Size of Community (Number of Users)	Location(s) of Community	Application(s) Used by Community
Headquarters	350	Portland	All
Office paper manufacturing and sales	200	Seattle	All
Newsprint, cartons, and boxes manufacturing and sales	250	Spokane	All
Wood pulp and chemicals manufacturing and sales	150	Boise	All
Other smaller manufacturing and sales offices	25–75	Western United States	All

Data Stores (Servers)

Table 11-5 shows the data stores that were identified at Klamath.

Table 11-5 Klamath Data Stores

Data Store	Location	Application(s)	Used by User Community (or Communities)
Mainframe	Portland	Manufacturing support system	All manufacturing sites
UNIX servers	Two in Portland	Financial modeling	Finance departments in Portland, Seattle, Spokane, and Boise
Windows servers	Portland, Seattle, Spokane, Boise	Sales order-entry and tracking system	All sales sites
AFP servers	Portland, Seattle, Spokane, Boise	Graphics production	Graphics departments in Portland, Seattle, Spokane, and Boise
Video server (new)	Portland	Distance learning	All

Current Network

The current WAN consists of dedicated 64-kbps data circuits that connect the 15 sites in a partial-mesh topology. Voice traffic is carried on separate 64-kbps circuits. A WAN service provider leases the 64-kbps lines to Klamath and also provides Internet access via a T1 circuit that connects a router at the Portland headquarters to a router at the provider’s site. The router at the Portland headquarters acts as a packet-filtering firewall. The core of the data network is a full mesh of 64-kbps circuits that connects the major sites. A router at each site connects Ethernet LANs to the WAN, as shown in Figure 11-5 .

Figure 11-5 Existing Core WAN at Klamath

Traffic Characteristics of the Existing WAN

As Klamath has grown over the years, network performance has degraded. Users report that the network is slow, especially during the busiest hour between 10 a.m. and 11 a.m. Users of the manufacturing support system report that it sometimes takes 2 or 3 minutes for their screens to unlock after they enter information. Users of the sales order-entry system and the financial modeling applications also report slow response times.

A WAN protocol analyzer was used at each of the major sites to measure current bandwidth usage on the 64-kbps data circuits. It was determined that every circuit in Portland was approaching saturation, with an average utilization of 80 percent in a 10-minute window. WAN circuits between Seattle and Spokane, Spokane and Boise, and Boise and Seattle were also heavily used, with an average utilization in a 10-minute window of 70 percent.

The protocol analyzer was also used to study protocol and traffic characteristics. The following conclusions were made:

· No single protocol was causing any serious problems.

· Although there were quite a few retransmissions, no applications appeared to retransmit too quickly.

· Applications appeared to have been optimized to use large frame sizes and large window sizes.

· Broadcast traffic accounted for about 5 percent of the network utilization and appeared to be normal routing and service-advertising packets.

· The average error rate for the circuits was one cyclic redundancy check (CRC) error per two million bytes of data (which is acceptable).

The status of the routers in the core of the network was also checked. The following Cisco IOS commands, which were discussed in Chapter 3 , “Characterizing the Existing Internetwork,” were used to check the routers:

· The show processes command indicated no problems with CPU overutilization.

· The show buffers command indicated no problems with buffers.

· The show interfaces command indicated that the routers were dropping frames from the output queue of the serial WAN ports at a rate of about 1 in 20 frames, or 5 percent. This appeared to be caused by too much network traffic destined for the 64-kbps circuits, and was considered a problem.

The end result of the analysis of the existing core WAN was that the core WAN was congested due to too much traffic caused by normal application behavior.

WAN Design for Klamath Paper Products

A decision table was used as part of the design process for Klamath. Klamath’s major goals were consolidated and critical goals were placed at the top of the table, as shown in Table 11-6 . Potential options were placed in the leftmost column and evaluated on whether they met a critical goal.

Note

If all options had met all critical goals, then other goals could have been listed also to further the decision-making process. Options could have been evaluated on how well they met noncritical goals on a scale from 1 to 10, as shown in Table 7-1 , “Example Decision Table,” in Chapter 7 . However, in the case of Klamath, it was not necessary to go beyond evaluating how well options met critical goals.

After analyzing business and technical goals, characterizing the existing core WAN, and analyzing the options available from the WAN service providers in the area, Klamath decided to update the core WAN architecture from 64-kbps lines to Metro Ethernet using an E-LAN service and MPLS transport. This choice will keep costs low and benefit from the QoS features of Metro Ethernet.

Table 11-6 WAN Technologies Decision Table

The three main contenders, besides Metro Ethernet, were ATM, Ethernet over ATM, and Frame Relay. ATM was ruled out because ATM router interfaces are too expensive. Ethernet over ATM was ruled out because the services providers in the area don’t offer it. Frame Relay was seriously considered, but Metro Ethernet was selected instead of Frame Relay because it was designed to handle environments with diverse applications, such as the delay-sensitive videoconferencing and interactive manufacturing-support and financial-modeling applications at Klamath.

In the new design, the delay-sensitive traffic will be carried on virtual circuits that are distinct from the circuits that carry other data. Traffic management parameters will be specified and implemented so that the network accepts and carries traffic according to the QoS needs of the different applications.

The current WAN service provider that leases the existing 64-kbps circuits was selected as the service provider for the new Metro Ethernet WAN. The current provider was selected because it offers the following advantages over other providers:

· A proven history of supplying highly reliable Metro Ethernet services to customers

· Excellent pricing for Metro Ethernet that was comparable in price to the cost of adding numerous 64-kbps circuits to meet capacity requirements

· The ability to allow Klamath to keep their current IP addressing scheme

· 24-hour support, 7-days a week, with a guaranteed MTTR of 1 hour

· A single point of contact who is responsible for Klamath’s service

Each site in the core of the network will connect to the network via a 10/100 Ethernet interface in a router. For now, 10-Mbps Ethernet will be used. Upgrading to 100 Mbps or 1000 Mbps will be possible in the future, but for now, 10-Mbps Ethernet was chosen because the service provider offers attractive pricing for 10-Mbps Metro Ethernet and because 10 Mbps is sufficient capacity for the current network applications.

Klamath will replace the existing routers in the core of the network with new high-end routers that provide superior packets-per-second throughput, high-availability features, and support for optimization features. The routers will support Ethernet interfaces for connectivity to the service provider’s Metro Ethernet and to the internal LANs and mainframe. Figure 11-6 shows the new design for Klamath’s core network.

Figure 11-6 New Core WAN at Klamath