Designing Your Converged Network

Step 3 Calculate the voice delay budget. For voice traffic, delays on the network can be particularly problematic. Determine how much delay, if any, your organization is willing to accept.

Step 4 Plan for capacity. Determine how many trunks and how much bandwidth is required to meet your needs as defined in the previous steps.

Step 5 Analyze the costs. Determine whether the costs of implementing, operating, and maintaining the network design you have specified are realistic now and in the future.

Auditing Your Network

The first step in designing a network is to evaluate what currently exists. Consider the following issues:

Equipment and infrastructure. Evaluate your network's hardware capabilities—PBXs, telephones, trunks (lines between PBXs or between a PBX and the telephone company), leased lines (dedicated, point-to-point lines), servers, routers, and other equipment.
Cost. Assess your existing networking costs.
Growth. Determine whether your network will continue to meet your organization's projected voice and data needs. Identify upcoming projects and determine their impact on the network.
Performance. Determine whether the level of service is acceptable, particularly for voice performance and quality.
Traffic patterns. Conduct a study of current traffic patterns. A traffic study can identify traffic bottlenecks and can help you identify areas of the network where bandwidth is underutilized.

Setting Network Objectives

Identify the key objectives of your network:

Determine the dominant traffic type (voice or data) that you need to carry on the network. Typically, the volume of data traffic exceeds that of voice traffic.
Set voice-quality objectives to help you design networks with acceptable amounts of delay and compression.
Determine the level of performance expected of the integrated network. For example, you need to plan for how the addition of voice can affect the network's ability to carry data traffic.
Set financial objectives to help you determine what return on investment you expect and over what period of time.

Calculating the Voice Delay Budget

The delay budget is the amount of delay that your organization is willing to tolerate in your planned network. Consider the following types of delay when calculating the delay budget:

Propagation delay is based on the total distance between source and destination. For planning purposes, use 6 microseconds per kilometer to allow for propagation delay.
Serialization delay is the time that it takes to place a packet on the network. The higher the circuit speed, the less the serialization delay. For example, it takes about 5 microseconds to send a 1024-byte packet over a 1.544-Mbps line. The type of compression being used has an important effect on serialization delay. For example, if voice data is compressed into 8-kbps streams (using CS-ACELP), the resulting delay should be about 20 microseconds.
Packetization delay is the process of holding digital voice samples until enough samples are collected to fill the packet. The size of the payload (quantity of data, in bytes) to be included in each packet is typically configurable, giving you control over packetization delay. Increased packetization delay results in increased serialization delay.
Queuing delay occurs when a packet has to wait because there are other packets ahead of it on the trunk. Because queuing delay is based on the volume of traffic, queuing delay increases as traffic increases. Another factor affecting queuing delay times is the size of the packet currently being processed.

The International Telecommunication Union (ITU) recommends that there be no more than 150 milliseconds (ms) of one-way delay. Take a combined score of all of the above delays and compare the result with this standard.

Note Delays ranging from 150 to 400 ms may be acceptable in some cases. For example, a 200-ms delay from Chicago to New York is unacceptable, but the same delay from Chicago to Singapore might be acceptable. Furthermore, higher delays can be acceptable if cost savings are taken into account.

Planning for Capacity

Capacity planning involves provisioning the number of lines and trunks necessary to achieve bandwidth goals. (For details on provisioning trunks, see "Traffic Engineering.")

Provisioning Trunks

When planning the number of system trunks that you need, the main considerations are traffic volume and flow, selected grade of service, and your objectives.

Based on your proposed network design and the required number of trunks between locations, you can calculate the required bandwidth. When calculating bandwidth, you will need to consider issues such as compression, network overhead, and network utilization.

Analyzing the Costs

Once you have conducted a network audit, set your objectives, computed the delay budget, and completed capacity planning, you are ready to do a complete financial analysis. Determine whether the projected costs of the network you intend to build are justified in relation to the benefits received, and calculate the expected return on investment.

The cost of your internetwork is much more than the sum of equipment purchase orders. You must consider the entire life cycle of your internetworking environment. A brief list of costs associated with internetworks follows:

Equipment hardware and software costs. Consider what you are really buying when you purchase your system—costs should include initial purchase and installation, maintenance, and projected upgrades.

Performance trade-off costs. Consider the cost of going from a 5-second response time to a 1/2-second response time. You can make improvements in areas such as media (different types of cables support Ethernet and Fast Ethernet, for example), network interfaces, internetworking nodes (device locations), modems, and WAN services.
Cable installation costs. The costs associated with installing cable may include labor, fees associated with local code conformance, and expenses related to environmental requirements (such as asbestos removal). Important cost-cutting elements include developing a well-planned wiring closet layout and implementing color code conventions for cable runs.
Expansion costs. Calculate the cost of removing and replacing existing media, adding functionality, or moving to a new location. Projecting your future requirements saves time and money.
Support costs. Complicated internetworks can be expensive to monitor, configure, and maintain. Your internetwork should be no more complicated than necessary. Costs may include training, information technology (IT) personnel (network managers and administrators), procuring spares, and replacing equipment. Additional costs can include those related to network management stations (servers) and power.
Cost of downtime. Evaluate the cost for every minute that a user is unable to access the telephone network, a file server, or a centralized database. If this cost is high, you must attribute a high cost to downtime. If the cost is high enough, fully redundant internetworks (duplicate systems for backup purposes) might be necessary.
Opportunity costs. The opportunity costs of not switching to newer technologies and topologies might be lost competitive advantage, lower productivity, and slower overall performance. Any effort to integrate opportunity costs into your analysis can help you make accurate comparisons at the beginning of your project.
Sunken costs. Your investments in existing cable plant, routers, switches, servers, and other equipment and software are your sunken costs. If the sunken cost is high, you might need to modify your networks so that your existing internetwork can continue to be used. Although comparatively low incremental costs might appear to be more attractive than significant redesign costs, your organization might pay more in the long run by not upgrading systems. Excessive reliance on sunken costs can cost your organization sales and market share when calculating the cost of internetwork modifications and additions.

Topologies for Simple Internetworks

A topology is the arrangement of the interconnected networks in a wide-area network (WAN). Numerous topologies are possible, each offering a different mix of cost, performance, and scalability (the degree to which future growth is possible). The following represent some of the simplest topologies:

Peer-to-peer topology. The simplest, least expensive way to interconnect a small number of sites.
Ring topology. Provides route redundancy and dynamic routing at a low cost, making it more robust than the peer-to-peer topology.
Star topology. Offers a more scalable solution than peer-to-peer and ring topologies, making it a popular choice for interconnecting multiple sites.
Partial meshed topology. The most complex and expensive of the four basic topologies, it avoids the single points of failure that sometimes occur in star topologies.

Each topology is examined in the rest of this section, including its relative cost, performance, scalability, and technology implications.

Peer-to-Peer Topology

A peer-to-peer WAN is a relatively simple way to interconnect a small number of sites using leased lines or virtually any other media. (WANs that comprise only two locations must be interconnected in this way.) Figure 3-1 depicts a small, peer-to-peer WAN.

Figure 3-1: Peer-to-Peer WAN Constructed with Leased Lines

The peer-to-peer topology is the least expensive solution for WANs that comprise a small number of internetworked sites. Because each site comprises, at most, one or two links to the rest of the network, static routing can be used.

Routers can route in a two basic ways. They can use preprogrammed static routes, or they can dynamically calculate routes using any one of a number of dynamic routing protocols. Routers use dynamic routing protocols to discover routes. Routers then forward packets over those routes.

Limitations

Static routing can take time to establish but avoids the network overheads of dynamic routing protocols. However, this lack of route redundancy is an important limitation of the peer-to-peer topology.

Peer-to-peer WANs have two other limitations. First, they do not scale very well. As sites are added to the WAN, the number of hops (devices, typically routers or switches) between any two sites tends to increase, resulting in inconsistent performance in cross-site communications. The actual degree to which performance varies depends on many factors, including the following:

Type and capacity (bandwidth) of media
Degree to which media are being used
Geographic distances between sites

The degree to which this lack of scalability affects your internetwork depends on how much you expect it to grow. If you do not expect your internetwork to grow beyond a few sites, peer-to-peer might be the appropriate topology.

The second limitation of this approach is its vulnerability to component failure. There is only a single path between any two sites. Consequently, an equipment or facility failure anywhere in a peer-to-peer WAN can split the WAN. Depending on the actual traffic flows and the type of routing implemented, this kind of failure can severely disrupt communications in the entire WAN.

Note See "Data Networking Fundamentals," for more information about routers, switches, routing protocols, and scalability.

Ring Topology

A peer-to-peer internetwork becomes a ring internetwork when you add a node (typically a router or PBX) and an extra port to two of the routers or PBXs, which provides route redundancy. The additional route enables small networks to implement dynamic routing protocols because traffic has an additional path over which it can travel. Because the cost of most media is affected by distance, you should design the ring to minimize distances. Figure 3-2 illustrates this WAN topology.

Figure 3-2: Ring-Shaped WAN

You can use a ring-shaped WAN constructed with leased lines to connect a small number of sites and provide route redundancy at a potentially small incremental cost. With redundant routes and a dynamic routing protocol, you have more flexibility than you do with static routing. For example, dynamic routing protocols can automatically detect and recover from adverse changes in the WAN operating condition by routing around degraded or non-operational links.

Limitations

Ring topologies have several limitations. First, depending on the geographic dispersion of the sites, adding an extra node to complete the ring could be cost-prohibitive.

A second limitation of rings is that they are not very scalable. Adding new sites to the WAN increases the number of hops required to access other sites in the ring. For example, in Figure 3-2, adding a new location (X) that is in geographic proximity to User Locations C and D would require terminating the circuit from location C to D. Two new circuits would have to be ordered to preserve the integrity of the ring: one running from C to X and the other from D to X.

The final limitation of a ring is its potential hop intensity. A ring is not a good way to minimize the number of hops—each interior gateway router on a ring is only adjacent to two other interior gateway routers. The number of hops to any other site depends on the way in which these sites are connected. A ring does offer route redundancy, but the hop count between any given source and destination address pair can vary widely.

The ring topology, given its limitations, is likely to be of value in connecting only a very small number of sites. It is preferable to the peer-to-peer topology only because it can provide a redundant path to the locations within the ring.

Star Topology

A variant of the peer-to-peer topology is the star topology, so named for its shape. A star is constructed by connecting all nodes (routers or PBXs) to a common node in the center, called a concentration node. The concentration node in a star topology can also be used to interconnect the LANs installed at that location with each other as well as with the WAN.

You can build a star topology with almost any dedicated media, including leased lines. Figure 3-3 shows an example of a star-shaped WAN.

Figure 3-3: Star-Shaped WAN

A star topology WAN with leased lines is much more scalable than a peer-to-peer or ring network. Adding locations to the star does not require re-engineering existing media. Only a new facility between the concentration node and the node at the new location is required.

The star topology improves upon the scalability of peer-to-peer networks by using a node to interconnect, or concentrate, all the other networked nodes. This scalability is achieved with a modest increase in the number of nodes, node ports, and media compared to a comparable peer-to-peer topology. Star topologies can actually be developed with fewer resources than ring topologies, as Figure 3-2 and Figure 3-3 demonstrate. The scalability of a star topology is limited by the number of ports that the concentration node can support. Expansion beyond its capacity requires re-engineering the topology into a multitiered topology or replacing the concentration node (if it is a low-end router) with a more robust device.

Another benefit of a star topology is improved network performance. Overall performance in a star topology is, in theory, always better than in a ring or peer-to-peer network because all network-connected devices are no more than three hops away from each other. These three hops are the node at the user's location, the concentration node, and the node at the destination. This degree of consistency is unique to the star topology.

Note In very small WANs, such as those with only two or three internetworked locations, it might be difficult to perceive any difference between a star topology and a peer-to-peer topology. The benefits of a star topology become increasingly apparent as your network increases in size.

Limitations

Star topologies have two problems; they are subject to single points of failure and there is no route redundancy. A single point of failure means that all WAN communications can be disrupted if the concentration node experiences a failure. The lack of route redundancy means that if the concentration node fails, you are out of service until that failure is rectified.

You can compensate for these limitations in a variety of ways: by implementing a slightly more complex topology, such as the partial meshed, or even by splitting the star into two smaller stars that are linked together. In the event of a failure, only half of the remote locations would be affected. Additionally, you can use dial-on-demand technologies such as Integrated Services Digital Network (ISDN) to reestablish a limited amount of communications.

Partial Meshed Topology

Partial meshed topologies are highly flexible and can have very different configurations. In a partial meshed topology, the nodes are more closely integrated than in any other basic topology, but they are not fully interconnected. (Fully interconnected nodes form a fully meshed topology.) Figure 3-4 shows a partial meshed topology.

Figure 3-4: Partial Meshed Topology

A partial meshed WAN topology is readily identified by its almost complete interconnection of every node with every other node.

This topology can minimize hops for most of the WAN's users. Unlike fully meshed networks, a partial meshed network can reduce startup and operational expenses by not interconnecting low-traffic segments of the WAN, which makes it somewhat more scalable and affordable than a fully meshed topology.

General Network Design Guidelines

Good network design is based on the following key principles:

Eliminate single points of failure. Redundancy prevents a single failure from isolating any part of the network. You need to consider two aspects of redundancy: backup and load balancing. In the event of a failure in the network, an alternative or backup path should be available. Load balancing occurs when two or more paths to a destination exist and can be used according to the network load. The level of redundancy required varies from network to network.

Characterize application and protocol traffic. The flow of application data affects client-server interaction and is crucial for efficient resource allocation, such as the number of clients using a particular server or the number of client workstations on a network segment.

Analyze bandwidth availability. Plan so that each network layer has approximately the same amount of bandwidth available to it.

Build networks using a hierarchical model. A hierarchical model allows autonomous segments to be internetworked. (See "The Hierarchical Model" in the next section.)

Note See "Data Networking Fundamentals," for a more detailed explanation of data networking terms and concepts.

The Hierarchical Model

Figure 3-5 shows a general view of a hierarchical network design. A hierarchical network has three layers—core, distribution, and access.

Each layer of the hierarchical model has a specific role:

The core layer uses high-speed switching to provide optimal transport between sites. The Cisco ICS 7750 is designed to be used with a configurable number of Catalyst 3524-PWR XL switches.
The distribution layer provides policy-based connectivity through security, load balancing, and routing services, and connects the core layer to the access layer. The Cisco ICS 7750 provides software solutions including Cisco ICS 7700 System Manager and Cisco CallManager 3.0.
The access layer is the point at which local users are allowed into the network. In a non-campus environment, the access layer can give remote sites access to the corporate network via WAN technologies such as ISDN or leased lines.

: For example, the access layer could consist of multiple Cisco ICS 7750s, possibly used with multiple routers or other devices at the edge of one or more campus networks. In a campus network the access layer might require only a single Cisco ICS 7750.

Figure 3-5: Hierarchical Network Model

Why Use a Hierarchical Model?

Some of the main reasons to use a hierarchical model are listed below:

Cost reduction. You can purchase the appropriate internetworking devices for each layer of the hierarchy, thus avoiding spending money on unnecessary features. In a hierarchical network you can distribute management systems and responsibility to different layers to control management costs. Also, modularity helps you accurately plan for capacity within each layer, reducing wasted bandwidth.
Simplification. Modularity helps you ensure that each part of your design is easy to understand, minimizing the need for training network operations personnel and expediting the implementation of your design. Testing a network design is easier because there is clear functionality at each layer. Fault isolation is improved because network technicians can recognize the transition points in the network.
Extensibility. The cost of upgrading is limited to a small subset of the overall network. In large flat or meshed network architectures, replacing one device can affect numerous subnetworks because of the complex interconnections.
Scalability. You can replicate the modular elements of a hierarchical network as the network grows. Because each instance of a module is consistent, expansion is easy to plan and implement. For example, planning a campus network for a new site might simply be a matter of replicating an existing campus network design.

Tips Today's fast-converging routing protocols are designed for hierarchical topologies. To control routing CPU overhead and bandwidth consumption, use modular hierarchical topologies with such protocols as Open Shortest Path First (OSPF), Border Gateway Protocol (BGP), and Enhanced Interior Gateway Routing Protocol (Enhanced IGRP).

Note See "Data Networking Fundamentals," for a more detailed explanation of routers, switches, and routing protocols and for an in-depth discussion of design issues related to fault tolerance and scalability.

Designing the Core Layer

The core layer of a three-layer hierarchical topology is the high-speed backbone of the internetwork. Because the core layer is critical for interconnectivity, you should design the core layer with redundant components.

When configuring nodes in the core layer, use routing features that optimize packet throughput. Avoid using packet filters (see "Data Networking Fundamentals") or other features that slow down the manipulation of packets. You should optimize the core for low latency and good manageability.

The core should have a limited and consistent diameter. You can add distribution-layer nodes and client LANs to the model without increasing the diameter of the core. Limiting the diameter of the core provides predictable performance and easy troubleshooting.

If you need to connect to other enterprises via the Internet, the core topology should include one or more links to external networks. By centralizing these functions in the core layer you can reduce the potential for routing problems, and by keeping Internet connections to a minimum, you can reduce security concerns.

Designing the Distribution Layer

The distribution layer lies between the access and core layers of the network. The distribution layer has many roles, including controlling access to resources for security reasons and controlling network traffic that traverses the core for performance reasons. The distribution layer often delineates broadcast domains (sets of devices that receive broadcasts from any other device in the set), although you can define broadcast domains at the access layer as well.

To improve routing protocol performance, the distribution layer can summarize routes from the access layer. For some networks, the distribution layer offers a default route to access-layer routers and runs only dynamic routing protocols when communicating with core nodes.

Another function of the distribution layer is address translation. With address translation, devices in the access layer can use private addresses. The address-translation function converts the private addresses to legitimate Internet addresses for packets that traverse the rest of the organization's internetwork or the Internet.

Designing the Access Layer

The access layer gives end users access to the internetwork. Nodes are implemented at the access layer in campus networks to meet the demands of applications that need a considerable amount of bandwidth or cannot withstand the variable delay characteristic of shared bandwidth.

For internetworks that include small branch offices and telecommuter home offices, the access layer can provide access into the corporate internetwork using wide-area technologies such as ISDN, leased lines, and analog modem lines. You can implement routing features such as dial-on-demand (DDR) routing and static routing to control bandwidth utilization and minimize cost on access layer remote links. (DDR keeps a link inactive except when specified traffic needs to be sent.)

Prioritizing Network Design Goals

You will have to make difficult decisions when budgeting for and building (or upgrading) your internetwork. The following are examples of typical design decisions:

Availability versus cost. To meet high expectations for availability (the percentage of time that a system is operating), redundant components are often necessary, which raises the cost of a network implementation. Conversely, the availability rate can decrease when aggressive cost-cutting measures result in reduced IT training and staffing.
Performance versus cost. To meet rigorous performance requirements (particularly for voice traffic), high-cost media (such as fiber-optic cable) and equipment (such as PBXs, servers, and high-end routers) may be required.
Scalability versus availability. When implementing a scalable network, the availability rate may decrease because a scalable network is always in flux as new users and sites are added.
Security versus ease of use. To enforce strict security policies, expensive monitoring might be required and users might have to sacrifice some ease of use.

Try to prioritize your most important network design goals. Prioritizing should help you get through this difficult decision-making process.

The most important goals typically include the following:

Affordability. The cost-effectiveness of the network design.
Scalability. The ability of the network design to accommodate future requirements without extensive modification.
Availability. The percentage of time that the network is operational.
Performance. The ability of the network to perform at acceptable rates, even under heavy traffic conditions.
Security. The ability of the network design to protect sensitive information and to prevent unauthorized individuals from gaining access to the network.
Manageability. The simplification of network configuration, operation, and monitoring through software tools and utilities.
Usability. The simplification of network access for end users through software applications such as unified messaging.
Adaptability. The ability of the network to handle changing traffic patterns and isolate problem areas.

Note Scalability, availability, and security are discussed in more detail in "Data Networking Fundamentals."

Keep in mind that sometimes making design decisions is more complex than what has been described because goals can differ for various parts of an internetwork. One group of users might value availability more than affordability. Another group might deploy state-of-the-art applications and value performance more than availability. In addition, sometimes a particular group's goals are different from the goals for the internetwork as a whole. If this is the case, document individual group goals as well as goals for the entire internetwork. Later you might be able to implement LAN technologies that meet individual group goals and WAN technologies that meet overall goals.

Table of Contents

Designing Your Converged Network