Chapter 9 Network Design Networking in the Internet Age by Alan Dennis 1 Copyright © 2002 John Wiley & Sons, Inc. Copyright John Wiley & Sons, Inc. All rights reserved. Reproduction or translation of this work beyond that named in Section 117 of the United States Copyright Act without the express written consent of the copyright owner is unlawful. Requests for further information should be addressed to the Permissions Department, John Wiley & Sons, Inc. Adopters of the textbook are granted permission to make back-up copies for their own use only, to make copies for distribution to students of the course the textbook is used in, and to modify this material to best suit their instructional needs. Under no circumstances can copies be made for resale. The Publisher assumes no responsibility for errors, omissions, or damages, caused by the use of these programs or from the use of the information contained herein. 2 Chapter 9. Learning Objectives • Be familiar with the overall process of design and implementing a network • Be familiar with techniques for developing a logical network design • Be familiar with techniques for developing a physical network design • Be familiar with network design principles 3 Chapter 9. Outline • Introduction – Traditional Network Design, Building Block Network Design • Needs Analysis – Geographic Scope, Application Systems, Network Users, Categorizing Network Needs, Deliverables • Technology Design – Designing Clients and Servers, Designing Circuits and Devices, Network Design Tools, Deliverables • Cost Assessment – Request for Proposal, Selling the Proposal to Management, Deliverables • Designing for Network Performance – Managed Networks, Network Circuits, Network Devices, Minimizing Network Traffic 4 Introduction 5 Traditional Network Design • The traditional network design approach follows a structured systems analysis and design process similar to that used to build application systems. – The network analyst meets with users to determine the needs and applications. – The analyst estimates data traffic on each part of the network. – The analyst designs circuits needed to support this traffic and obtains cost estimates. – Finally, a year or two later, the network is implemented. 6 Traditional Network Design • Three forces are making the traditional design approach less appropriate for many of today’s networks: – 1. The underlying technologies used in computers, networking devices and network circuits are rapidly changing. – 2. Network traffic is growing rapidly. – 3. The balance of costs has changed dramatically over the last 10 years. 7 Building Block Network Design (Figure 9-1) • While some organizations still use the traditional approach, many others use a simpler approach to network design, the building block approach. • This approach involves three phases: needs analysis, technology design, and cost assessment. • When the cost assessment is initially completed, the design process returns to the needs analysis phase and cycles through all three phases again, refining the outcome of each phase. • The process of cycling through all three design phases is repeated until a final design is decided on (Figure 9-2). 8 Figure 9-1 Network Design 9 Fig. 9-2 The cyclical nature of network design 10 Needs Analysis 11 Needs Analysis • The first step is to analyze the needs of network users along with the requirements of network applications. • Most efforts today involve upgrades and not new network designs, so most needs may already be understood. • LAN and BN design issues include improving performance, upgrading or replacing unreliable or aging equipment, or standardizing network components to simplify network management. • At the MAN/WAN level, circuits are leased and upgrades involve determining if capacity increases are needed. • The object of needs analysis is to produce a logical network design, which describes what network elements will be needed to meet the organization’s needs. 12 Geographic Scope (Figure 9-3) • Needs analysis begins by breaking the network into three layers based on their geographic and logical scope: – The access layer which lies closest to the user – The distribution layer which connects the access layer to the rest of the network – The core layer which connects the different parts of the distribution layer together. 13 Figure 9-3 Geographic Scope 14 Application Systems • The designers must review the applications currently used on the network and identify their location so they can be connected to the planned network (baselining). • Next, applications expected to be added to the network are included. • It is also helpful to identify the hardware and software requirements and protocol type for each application. 15 Network Users • In the past, application systems accounted for the majority of network traffic. Today, much network traffic comes from Internet use (i.e., e-mail and WWW). • The number and type of users that will generate network traffic may thus need to be reassessed. • Future network upgrades will require understanding how the use of new applications, such as video, will effect network traffic. 16 Categorizing Network Needs • The next step is to assess the traffic generated in each segment, based on an estimate of the relative magnitude of network needs (i.e. typical vs. high volume). This can be problematic, but the goal is a relative understanding of network needs. • Once identified, network requirements should be organized into mandatory requirements, desirable requirements, and wish list requirements. 17 Deliverables • The key deliverable for the needs assessment stage is a set of network maps, showing the applications and the circuits, clients, and severs in the proposed network, categorized as “typical” or “high volume”. 18 Figure 9-4 Sample needs assessment 19 Technology Design 20 Technology Design • After needs assessment has been completed, the next design phase is to develop a technology design (or set of possible designs) for the network. 21 Designing Clients and Servers • In the building block approach, the technology design is specified by using “standard” computer units: – “Typical” users are allocated “base level” client computers, as are servers supporting “typical” applications. – “High volume” users and servers are assigned “advanced” computers. – The definition for a standard unit, however, keeps changing as hardware costs continue to fall. 22 Designing Circuits and Devices • Two interrelated decisions in designing network circuits and devices are: 1) deciding on the fundamental technology and protocols and 2) choosing the capacity each circuit will operate at. • Capacity planning means estimating the size and type of the “standard” and “advanced” network circuits for each type of network. • This requires some assessment of the current and future circuit loading in terms of average vs. peak circuit traffic. 23 Estimating Circuit Traffic • The designer often starts with the total characters transmitted per day per circuit, or if possible, the maximum number of characters transmitted per two second interval if peak demand must be met. • While no organization wants to overbuild its network and pay for unneeded capacity, going back and upgrading a network often significantly increases costs. 24 Network Design Tools • Network modeling and design tools can perform a number of functions to help in the technology design process. • Some modeling tools require the user to create the network map from scratch. Other tools can “discover” the existing network. • Once the map is complete, the next step is to add information about the expected network traffic and see if the network can support the level of traffic that is expected. This may be accomplished through simulation models. • Once simulation is complete, the user can examine the results to see the estimated response times and throughput. 25 Deliverables • The key deliverables at this point are a revised set of network maps that include general specifications for the hardware and software required. • In most cases the crucial issue is the design of the network circuits. 26 Figure 9-5 Physical Network Design 27 Cost Assessment 28 Cost Assessment • Cost assessment’s goal is to assess the costs of various network alternatives produced as part of technology design. Costs to consider include: Circuit costs for both leased circuits and cabling. Internetworking devices such as switches and routers. Hardware costs including servers, memory, NICs & UPSs. Software costs for operating systems, application software and middleware. Network management costs including special hardware, software, and training. Test and maintenance costs for monitoring equipment and supporting onsite repairs. Operations costs to run the network. 29 Request for Proposal (RFP) • While some components can be purchased “offthe-shelf”, most organizations develop an RFP before making large network purchases. • The RFP creates a competitive environment for providing network equipment and services (see Figure 9-6). • Once vendors have submitted network proposals, the organization evaluates them against specific criteria and selects the winner(s). • Multi-vendor selections have the advantage of maintaining alternative equipment and services sources, but are also more difficult to manage. 30 Figure 9-6. Request for Proposal • Background Information – Organizational profile; Overview of current network; Overview of new network; Goals of the new network • Network Requirements – Choice sets of possible network designs (hardware, software, circuits); Mandatory, desirable, and wish list items, Security and control requirements; Response time requirements; Guidelines for proposing new network designs • Service Requirements – Implementation time plan; Training courses and materials; Support services (e.g., spare parts on site); Reliability and performance guarantees • Bidding Process – Time schedule for the bidding process; Ground rules; Bid evaluation criteria; Availability of additional information • Information Required from Vendor – Vendor corporate profile; Experience with similar networks; Hardware and software benchmarks; Reference list 31 Selling the Proposal to Management • An important hurdle to clear in network design is obtaining the support of senior management. • Gaining acceptance from senior management lies in speaking their language and presenting the design in terms of easily understandable issues. • Rather than focusing on technical issues such as upgrading to gigabit Ethernet, it is better to make a business case by focusing on organizational needs and goals such as comparing the growth in network use with the growth in the network budget. 32 Deliverables • There are three key deliverables for this step: 1. An RFP issued to potential vendors. 2. After the vendor has been selected, the revised set of network maps including the final technology design, complete with selected components. 3. The business case written to support the network design, expressed in terms of business objectives. 33 Designing for Network Performance 34 Network Management Software • Network management software is designed to provide automated support for some or all of the network management functions (Figure 9-8 shows an example). • There are three fundamentally different types of network management software: – Device management software – System management software – Application management software 35 Figure 9-8 Network management software (Source: HP OpenView) 36 Network Management Software • One major problem is ensuring that hardware devices from different vendors can understand and respond to the messages sent by the network management software of other vendors. • The two most commonly used network management protocols are: – Simple Network Management Protocol (SNMP, part of the TCP/IP protocol suite) – Common Management Interface Protocol (CMIP, developed by ISO) 37 Simple Network Management Protocol (See Figure 9-9) • SNMP: TCP/IP suite protocol for network management that allows agents to communicate with each other and other network devices • Agents: programs residing on network devices that gather and share network status information • Management Information Bases (MIBs): databases of network status statistics such as traffic levels, error rates & data rates • Network Management Console: when requested, data from the MIBs is sent to a Network Management Console. 38 Fig. 9-9 Network management with SNMP 39 Policy-based Management • In policy-based management, the network manager uses special software to set priority policies for network traffic. • These take effect when the network becomes busy. • For example, videoconferencing might be given a high priority since delays will have the highest impact on the performance of that application. 40 Traffic Analysis • The easy way to manage network traffic growth is simply to upgrade heavily used circuits. • A more sophisticated approach is to do traffic analysis. Consider the network in Figure 9-10: – The Toronto-Dallas network segment is heavily used (thick line), but the traffic is mostly moving between LA and NY. – The solution is to create a new LA to NY segment (dashed line). 41 Figure 9-10 Simple WAN 42 Service Level Agreements • More organizations establish service level agreements (SLAs) with common carriers. • SLAs specify the type of performance and fault conditions for their leased circuits. • For example, a 99.9% availability means the circuit will be down for 8.76 hours/year. • The SLA also often includes the maximum allowable response time. 43 Network Devices • Since network devices vary in their characteristics, a network’s performance will be influenced by the devices selected to operate on it. • Three important factors to network performance that are related to network device characteristics are: – Device latency – Device memory – Load balancing 44 Device Latency • Latency is the delay (waiting time) that occurs when a device processes a message. • Slow speed devices have high latency, while high speed devices have low latency. • The fastest devices run at wire speed. • Latency becomes a critical issue under high traffic conditions since high latency devices can create traffic congestion. • This is similar to the way that long lines of traffic form at tollbooths on highways during rush hour. 45 Device Memory • Memory and latency are related, since any device that operates at less than wire speed may need to store newly arrived packets. • Otherwise packets will need to be retransmitted, making the traffic situation worse. • Memory is also important for servers since memory access speeds are many times faster than hard disk access times. • The larger the memory a server has, the more likely it is able to process a request quickly, so Web and file servers should have the greatest amount of memory practical. 46 Load Balancing • Load balancing means sharing the processing load between servers. • A separate load balancing server is usually needed to allocate the work between processors. • The load-balancing server then allocates tasks to the other processors, using an algorithm such as a round robin formula. • An example of this is shown in Figure 9-11. 47 Figure 9-11 Network with load balancer 48 Minimizing Network Traffic • An alternative way of improving network performance is to minimize network traffic. • This can be done by shifting some of the data so it resides closer to the users (for example, a mirrored web site). • Two current approaches to minimize traffic flow are content caching and content delivery. 49 Content Caching (Figure 9-12) • Content caching means storing frequently used web pages locally, using a cache engine. • Web requests do not go out directly, they are first shunted by the router to the cache engine to see if they are available locally. • Traffic volume is lowered since many frequently requested web sites, such as yahoo.com, can be retrieved from the cache and don’t need to go out on the Internet. 50 Figure 9-12 Network with cache engine 51 Content Delivery (see Figure 9-13) • Another way to minimize network traffic is for web site operators to move content closer to users, called content delivery, is done by operating web servers near NAPs, MAEs and other exchanges to minimize network traffic. • If a web page of one of the content deliverers client’s is accessed, it checks if any web page components are located on a server near the requesting computer and sends those. • This benefits both the Web provider by lessening demand on its Web servers, the ISP by lowering demand on its Internet circuits, as well as decreasing Internet traffic overall. 52 Figure 9-13 Network with content delivery 53 End of Chapter 9 54