There are many different types of switching technologies that you may come across in educational and corporate networks. Some are becoming outdated, for example process switching is still extremely common in older hardware but involves a high level of overhead and latency. Newer routers have upgraded technology which has enabled their throughput substantially. The interface can be managed by the command console as normal ip address. Most of the newer switching technologies are based on the concept of caching routing information. This means that the processor in the router doesn’t need to be used to make all the routing decisions which significantly increases the speed and reduces load on the processor.
The amount of steps which are reduced due to the caching obviously varies depending on the content and the location of the cache. Most of the fast switching technologies have different forms and location of the cache. One of the fastest of the current solution is called silicon switching which is faster than other technologies such as autonomous switching simply because the cache is located much closer to the processors of the router’s interface.
Summary of Fast Switching Processes
- Initial Packet arrives on the interface processor hardware buffer
- Copied over to another packet buffer on the switch processor
- Silicon switching cache is checked for a corresponding entry
- Autonomous switching cache checked for corresponding entry
- Route processor checks fast switching cache for entry and will build appropriate encapsulation for destination interface. This information is received from fast cache in interrupt mode.
- New encapsulation copied across system bus to switch processor.
- Switch processor places oacket in output queue for the processor.
- Packet is copied across CxBus to a hardware buffer on the appropriate interface buffer and then to the destination interface.
The process is similar in the other two major technologies – autonomous switching and silicon switching. If you have even newer hardware, possibly the Cisco 7500 series and above you will have access to optimum and distributed switching which are potentially even faster. Optimum switching processes is effectively just an optimized version of fast switching. These switches have a new feature called VIP (versatile interface processor) which enables distributed switching. The switching process can be offloaded to this card when routing is performed across multiple interfaces using the same VPN card. The facility can be turned on and off by using the ip route cache distributed command.