The Distribution layer defines the boundaries for the network and provides packet manipulation of the network traffic. It aids in providing isolation from topology changes such as media translations, defining broadcast domains, QoS, security, managing the size of the routing table, aggregating network addresses, static route distribution, dynamic route redistribution, remote site connectivity, and inter−domain traffic redistribution.

Core Layer

The Core layer is designed to do one thing and one thing only: It switches packets at the fastest possible speed, providing the final aggregation point for the entire network. The devices at this layer must be fast and reliable. They should contain the fastest processors in the network. Connections at the Core layer must be of the highest possible bandwidth.

The Core layer makes no decisions about packet filtering or policy routing for two basic reasons. First, any filtering or policy decisions at this layer will add to the processing requirements of the system, thereby introducing latency in forwarding packets. Second, any forwarding mistakes at this level will severely impact the rest of the network.

Devices placed in the Core layer should be able to reach any device in the network. This doesn’t mean that they have to have a physical link directly to each device, but all devices must be reachable in the routing table. To prevent Core layer devices from having a path to every device in their routing tables, you should use network route summarization, which means defining the available routes for data traffic. If the Core layer is poorly designed, network instability can easily develop due to the demands placed on the network at this layer. A good tool in your toolbox to determine some of the problems in your network is Remote Monitoring.

Remote Network Monitoring

Remote Monitoring (RMON) is an industry−standard method used to monitor statistics on a network using Simple Network Management Protocol (SNMP). RMON allows a network administrator to obtain information about a switch’s Layer 1 or Layer 2 statistics. This type of information cannot be obtained by using the console port of the switch.

RMON collects information regarding connections, performance, configuration, and other pertinent statistics. Once RMON is configured on the switch, it runs continuously even when no clients are checking statistics. In fact, communication with an SNMP management station is not necessary. RMON can be configured to send trap messages to notify a management station when an error condition occurs that exceeds a currently configured maximum threshold.

With IP, nine different groups can provide RMON information. Four can be configured to provide information on a switch without an external device, such as a Switched Port Analyzer (SPAN). Cisco Catalyst switches support RMON information for IP traffic for the following four groups:

∙Statistics Group—Maintains utilization and error statistics. This group monitors collisions, oversized packets, undersized packets, network jabber, packet fragmentation, multicast, and unicast bandwidth utilization.

∙History Group—Provides periodical statistical information such as bandwidth utilization, frame counts, and error counts. The data can be stored for later use.

∙Alarm Group—Allows you to configure thresholds for alarms and the intervals at which to check statistics. Any monitored event can be set to send the management station a trap message regarding an absolute or relative value or threshold.

∙Event Group—Monitors log events on the switches. This group also sends trap messages to the management station with the time and date of the logged event, allowing the management station to create customized reports based on the Alarm Group’s thresholds. Reports can be printed or logged for future use.

33