Cisco Systems manufactures a large, varied, and ever-changing range of equipment. Over the years, the acquisition of a number of companies producing switches has meant that the range has not always appeared entirely consistent, but as time marches on, some of the differences in the underlying basics of the equipment are beginning to disappear. The most obvious differences in switch models now comes down to two factors: Are the switches modular (4000, 6500) or fixed footprint (2950, 3550), and do they support just layer 2 (2950) or can you buy a layer 3 capability (4000, 6500, 3550)?
Of course, the next question that arises is “Which switch should I choose?” Naturally there are issues of cost and size (in terms of ports and so on), but that may not be sufficient to help you design a complex network. So Cisco has pioneered some design guidelines that will help you put a specific Cisco box into a “location” in your internetwork, dependent upon the technologies required at that network point.
In order to understand all of this, there are two specific areas that we had to focus on. This first was how Cisco defines the network design model, in terms of redundancy, QoS, throughput, security, and so on, and how the Cisco models explain that to us. Cisco uses a three-layer model in which the access layer is used to provide redundant access to end users, the distribution layer manages policy, and the core layer provides fast access to the network backbone. Cisco also has a second model, related to its Secure Blueprint for Enterprise Networks (SAFE) guidelines, called the Enterprise Composite Module, which allows easy identification of modules such as the Management, Campus, Enterprise Edge, and SP Edge modules.
The second area we focused on was what technologies are available. Switches have traditionally been layer 2 devices, operating by forwarding data using MAC address tables. This is fast, but not very scalable, which means that routers, operating at layer 3, have been used. Modern devices can commonly combine the switching and routing processes, resulting in layer 3 switching. Layer 4 switching is an extension of that process, using the port fields inside TCP and UDP to assist with forwarding decisions. The total effect is commonly referred to as Multi- Layer Switching—MLS.