Handbook of Local Area Networks, 1998 Edition:LAN-based Application Development Issues and Solutions Click Here! Search the site:   ITLibrary ITKnowledge EXPERT SEARCH Programming Languages Databases Security Web Services Network Services Middleware Components Operating Systems User Interfaces Groupware & Collaboration Content Management Productivity Applications Hardware Fun & Games EarthWeb sites Crossnodes Datamation Developer.com DICE EarthWeb.com EarthWeb Direct ERP Hub Gamelan GoCertify.com HTMLGoodies Intranet Journal IT Knowledge IT Library JavaGoodies JARS JavaScripts.com open source IT RoadCoders Y2K Info Previous Table of Contents Next Windows Deployment A major pharmaceutical firm is converting its users from DOS to Windows. A corporatewide deployment standard is needed. Two approaches are possible: keep the applications on a central file server or load all the applications directly to the client hard disk. The former is much easier to manage, but may result in excess network traffic. Windows generates traffic as processes are swapped from RAM to hard disk. In addition, an application is never fully loaded into RAM when it is run Instead, dynamic linked libraries (DLLs) are swapped in and out of RAM as they are needed . The question is, how much traffic does this activity generate? How much network bandwidth will be washed on Windows overhead traffic? A simulation model was constructed to answer these questions. A group of five Windows power users was identified. Windows applications—a typical office suite—were loaded on a central server. Traffic to and from these users and the central application server was monitored for several days using an HP Network Advisor. From this data, a model was created for application traffic generated by a single user. This model was the used in the simulation to scale the number of users and determine the Windows overhead traffic that would be generated. A sample result for the network traffic is shown in Exhibit 6-5-3. The Windows application traffic for a 70-user Ethernet segment generated an average utilization of only 3.6%. An occasional peak near 20% is also seen in this trace. Exhibit 6-5-3.  Windows Application Traffic For Single Ethernet Segment with 70 Users The conclusion in this case is that the office suite application traffic would not generate significant overhead traffic. The desired scheme of central administration of network applications was compatible with efficient use of existing network capacity. Branch Office Network Design A major financial institution is upgrading the desktop applications suite offered to its customer representatives in the branch offices. These applications include customer account management, market data services, document imaging capability, and E-mail, among others. Older legacy Systems Network Architecture (SNA) terminal applications will also be included. The institution decided to use frame relay service to connect the branches with New York City headquarters. The question is, what bandwidth is needed for the access lines and permanent virtual circuits (PVCs)? Too many offices were involved to have a slow rollout of the new services. In addition, some of the new desktop applications were not yet completed, so testing was not practical. Nonetheless, guidance was needed and budget planning was essential given the size of the project. Simulating a Frame Relay Network Simulation was used to determine the characteristics of the frame relay cloud that would support the application suite. A schematic of the network is shown in Exhibit 6-5-4. Exhibit 6-5-4.  Frame Relay Branch Office Network •  Number of users in each branch office. •  Cloud access speed. •  PVC speed. •  Cloud error rates. •  Transport window size. •  Cloud delay characteristics. •  Router priority queuing. Based on these simulations, optimum choices can be made to maximize end-user performance and minimize network costs. Exhibits 6-5-5 and 6-5-6 illustrate the type of information that the simulations produce. Exhibit 6-5-5a shows the response-time characteristics for transfer of a 100KB document over a 56K-bps access line. This corresponds to about two 8 H by 11 pages of scanned text. The worst-case delay of more than three m inutes is clearly unacceptable. The priority queuing allowed the interactive traffic to move across the network efficiently, as shown in Exhibit 6-5-5b. The effect of going to T1 access is shown in Exhibit 6-5-6. Here, the maximum delay for the imaged document is only 1.5 seconds. The use of simulation allowed the end user to make intelligent tradeoffs between network cost and application performance. Exhibit 6-5-5.  Response Time For Interactive and Batch Processes Exhibit 6-5-6.  Delay from hq_server with T1 Access SUMMARY One of the major trends in networking today is quality of service. Now that most users are networked, they are demanding performance that meets their needs, and they are not willing to pay more for it. To manage this environment, network and technology managers should find great assistance in using simulation techniques. Close collaboration between end users and technology managers and effective use of network monitoring tools and the predictive power of simulation engines can help everyone find better solutions to today’s sophisticated IS architecture and design problems. Previous Table of Contents Next Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.