Requirements analysis is the first and most important step in the network designo The network design is the product, and the user is the customer. The network is built for the customers, and it must conform to their needs, wants, and desires. No single aspect of the design is more important than fully understanding the users' needs, for they ultimatelydictate the technology, protocols, hardware, software, and resources devoted to both access and backbone designo The two major views of requirements are those of the user and those of the designer. The user looks at the network from the outside (premises) in, and the designer looks at it from the inside (backbone network) out, thus creating two myopic views which must merge to provide a comprehensivecomplementary analysis beyond simple network ingress and egress designo Many questions must be asked before beginning the designo Both parties must "get to know each other" to begin the "marriage" of user and This working relationship is essential to the success of the designer. network's design and use. Changes and inaccuracies in user requirements can have devastating effects on both the accessand backbone network designo While user requirements change, the designer needs to set a time limit when the addition of any new requirements is frozen and the current ones analyzed. This chapter will examine how to look at the requirements from both the user's and the network design engineer or manager's perspective. But, first, we explore the many aspects of user requirements that set the stagefor the traffic-analysis and capacity-planning phases.
16.1. USER REQUIREMENTS THE USER'S VIEWPOINT How often do users proclaim "1 want it all now"? Users want as much network functionality as they can geto Bandwidth requirements are exploding as high-speed LANs and MANs proliferate and applications move toward distributed computing platformso Users who once carried on low-speed 629
terminal emulation sessions across the WAN are now more than ever accessing and sharing multimegabit data files across the WAN. Users who were once content to read e-mail messages are now scanning the Internet and retrieving huge text files with graphic images that cause the amount of data transmitted across their LAN to their workstationto increase by orders of magnitude. WAN link speeds of 9600 bps have yielded to speeds of 56 kbps, 1.544 Mbps, and 45 Mbps. Multiuser 4 Mbps and 10 Mbps LANs are yielding to single-user 100 Mbps LANs. But as these communications speeds increase, so does the increasing tangle of protocols, architectures, support and management systems, and a multitude of other factors influencing and controllingthe data flow between users. The line between network access devices and those once dedicated only to LAN communications has blurred and the two functions have merged. This is an exciting age, when users are taking more control of their data networks. Media and application enablers such as the Internet have only served to increase user interest in the WAN. Corporate and public communication datanetworks are playing catchup to the users demands. Networks where once the WAN bandwidths were the limiting factor are now facing protocol throughput problems that cannot be solved by simply adding more bandwidth in the WAN. As discussed in the beginning of the book, the user has become smarter and more sophisticated. Users no longer stand for the smoke-and-mirrors approach experienced withhierarchical MIS organizations in the pasto Historically, the MIS application backlog averaged from 1 to 2 years. Multimillion-dollar budgets for mainframes were tightly controlled and MIS was a mystery. These days, however, if corporate MIS do es not solve the connectivity problem in a timely manner, users buy their own routers, switching hubs, and workstations out of departmental budgets for a few...