The first step in the process is to identify what you're attempting to manage. The answer isn't as simple as "the network." In reality, the answer can vary dramatically from organization to organization, particularly since the responsibilities for different parts of the network are handled differently from one organization to the next. For example, a group responsible for the operation of the global corporate backbone WAN (perhaps with SLAs guaranteeing a level of service to business units) would have different baselining requirements than a group in a different organization that had end-to-end (NIC-to-NIC) responsibility for the network.

To understand the key service (or services) you must ask yourself what key services are being provided to customers (internal or external) by your part of the network organization, from the customer's perpective. Usually this will be a relatively small number of services--perhaps only one--but it is important to be specific in the definition of the service. A definition like "the global WAN backbone, including branch routers, excluding remote access and Internet connectivity" is better than "the corporate WAN."

Service Elements

The more thought you give to what the key service is from the user or customer perspective, the more clearly you will be able to determine the discrete elements that make up the service. In other words, figuring out all the links in the chain that make up the overall service. For example, in the case of our just described corporate WAN, the elements might include some or all of the following:

• Branch routers
• Backbone routers
• Public frame relay service
• Public ATM/Sonet service
• Frame relay switches
• ATM backbone switches
• Leased lines
• Managed Public IP Network (VPN)

Obviously other Network Managers with different kinds of responsibility will be providing different key services such as 'whole network NIC to NIC network connectivity' or 'Headquarters LAN infrastructure'. In these cases, there will be a number of additional LAN oriented service elements, for example:

• LAN Backbone Switches
• Campus routers
• Wiring closet concentrators (Hubs, switches, MAUs)
• Inter-switch backbone links (ATM, Fast/Gigabit Ethernet, FDDI)
• Server links (ATM, Fast/Gigabit Ethernet, FDDI, Ethernet, Token Ring)
• Workstation/Access LANs (Ethernet, Fast Ethernet, Token Ring)
• Individual VLANs, ELANs or PVCs

These lists are by no means complete. What I am trying to emphasize is that there are many different elements that go into delivering a particular key service to a customer. If the objective of baselining were simply service-level management, it would only be necessary to measure the overall quality of service for the key service as a whole. But for the purposes of performance management, it is necessary to identify which service elements affect the overall performance. This means analyzing a diagram of the network to determine which are the critical service elements. Obviously this will vary dramatically from network to network, but here are some general guidelines:

For leased line networks, critical service elements will be backbone links, links onto which other links are concentrated, and data center links.

For Frame Relay, public ATM networks and VPNs the critical service element will be the link(s) between the data center(s) and the network cloud.

For LANs the critical service elements will be the backbone links interconnecting hubs or switches as well as connections to major servers.