Cisco Catalyst - Interfaces

Interfaces

As Catalyst devices are primarily Ethernet switches, all modern Catalyst models have Ethernet interfaces, ranging from 10 Mbit/s to 10 Gbit/s depending on the model. Some models can accommodate Asynchronous Transfer Mode interfaces which can be used to bridge Ethernet traffic across wide area networks. Other models can support T1, E1, and ISDN PRI interfaces to provide connections to the PSTN. Legacy models supported a variety of interfaces, such as token ring, FDDI, and 100BaseVG, but are no longer sold by Cisco Systems.

Most models have basic layer 2 functions and are capable of switching Ethernet frames between ports. Commonly found additional features are VLANs (Cisco proprietary ISL or IEEE 802.1Q), trunking and QoS or CoS. The switches, whether IOS or CatOS, are fully manageable.

Many Catalysts that run IOS are also capable of functioning as a router, making them layer 3 devices; when coupled with TCP and UDP filtering, these switches are capable of layer 2-4 operation. Depending on the exact software image, a Catalyst that runs IOS may be able to tackle large-scale enterprise routing tasks, using router technologies like OSPF or BGP.

Most chassis-based Catalyst models have the concept of field-replaceable "supervisor" cards. These work by separating the line cards, chassis, and processing engine (mirroring most Cisco router designs). The chassis provides power and a high-speed backplane, the line cards provide interfaces to the network, and the processing engine moves packets, participates in routing protocols, etc. This gives several advantages:

• If a failure occurs, only the failed component needs to be replaced (typically a line card or supervisor). This means faster turnaround than having to uncable, unbolt, pull out, replace, re-bolt, and re-cable an entire switch, which may be as large as a quarter-rack, weigh over 150 pounds, and service over 500 cables.
• A redundant supervisor engine may be installed to rapidly recover from supervisor failures. This is subject to restrictions (as some switches don't support redundant supervisors), but typically results in restoration times under 90 seconds.
• A supervisor engine may be upgraded after purchase, increasing performance and adding features without losing any investment in the rest of the switch.

Additionally, most high-end switches off-load processing away from the supervisors, allowing line cards to switch traffic directly between ports on the same card without using any processing power or even touching the backplane. Naturally, this can't be done for all traffic, but basic layer-2 switching can usually be handled exclusively by the line card, and in many cases also more complex operations can be handled as well.