Managing a Network

Now that we have explored some of the various network management tools and protocols available, we next look at how these can be incorporated into a strategy for successfully managing a network, including service-level contracts (SLCs) and service-level agreements (SLAs).

Network Management Strategy

It is important to develop a network management strategy that details what information is to be collected from each device and how that information is to be analyzed. The appropriate protocols and tools, as described in the previous section, can then be selected.

Thresholds should be set so that alerts or alarms can be raised if parameters start to go out of range. To determine what these threshold levels should be, baseline measurements can be taken to create a snapshot of the "working" network. Instead of waiting for a failure and reacting to it, alerts and alarms relative to baseline measurements help the network manager to be proactive and solve problems before the network is adversely affected.

Cisco recommends the following network management best practices:^[6]

• Keep an archived copy of the software images (for example, Cisco IOS) and configurations of all devices.

• Keep an up-to-date inventory and log any configuration and software changes.

• Monitor critical parameters, including any syslog-reported errors, SNMP traps, and RMON statistics that are important for your network.

• Use tools to identify any configuration discrepancies (for example, CiscoWorks Campus Manager can detect inconsistent trunking configurations on switch ports).

SLCs and SLAs

SLCs and SLAs can also be a part of a network management strategy.

Key Point

An SLC specifies connectivity and performance levels for the service's end user, to be met by the service provider.

SLAs define specific service performance measurements between pairs of devices, for example, between a router and a server.

An SLC typically includes multiple SLAs, so a violation of any particular SLA could result in a violation of the overall SLC.

The service provider could be either within the organizationfor example, an IT department providing services to internal usersor an external company, such as an ISP providing hosted application services.

Traditional SLCs and SLAs concentrated on measurements over a specific network; for example, Frame Relay committed information rates. However, today's complex applications, including VoIP, for example, require end-to-end guarantees of service levels. For this type of environment, a new way of measuring and ensuring network performance is required. The Cisco IP SLAs technology, described in the next section, provides this functionality.

IP Service-Level Agreements

Cisco IOS IP SLAs^[7] technology is embedded in most IOS devices to allow users to analyze service levels being experienced by IP applications. IP SLAs allows test traffic to be generated and then measures the performance of that traffic through the network, either between Cisco IOS devices or from a Cisco IOS device to a remote device such as a server. Figure 9-5 illustrates IP SLAs operation.

Figure 9-5. IP SLAs Provides Network Performance Measurements

In Figure 9-5, the IP SLAs source generates some traffic destined for the application server, similar to an extended ping. The server's response includes a timestamp, which can then be used by the source to determine the network delay. The IP SLAs source also generates traffic destined for the IP SLAs responder device, which has IP SLAs software enabled. In this case, the response is more detailed, including, for example, the processing delays within the device.

Note

The Cisco IOS IP SLAs monitoring and management feature set replaced the Cisco IOS Service Assurance Agent (SAA) feature in Cisco IOS Release 12.3(14)T. ^[8] This feature set incorporates most of the features available in SAA, and SAA commands have been replaced by the IP SLAs monitor's command set and configuration structure.

The traffic generated by IP SLAs simulates real data by allowing many options to be configured, including source and destination IP addresses, UDP/TCP port numbers, ToS (either DSCP or IP precedence bits), VPN routing/forwarding instance (VRF), and Uniform Resource Locator (URL) web address. IP SLAs can be configured end to end over an entire network to best represent the metrics that an end user is likely to experience.

Measurements taken can include response time, latency, jitter, packet loss, voice quality scoring, network resource availability, application performance, and server response time.

IP SLAs is accessible using the command-line interface or SNMP. Performance-monitoring applications such as CiscoWorks IPM and other third-party Cisco partner performance management products can also use SNMP to configure and retrieve information from IP SLAs.