Layer 2 and layer 3 Switches

The first time I heard the term Layer 3 switch, I had a really hard time with it. It’s not that I couldn’t conceive what the device does, or how it would be implemented, but rather it was more an issue of simple pedantry. The term layer 3 switch is indeed slightly paradoxical at best and to some, completely misnamed.

Anyone with an understanding of the OSI model will already be nodding his head in agreement. It is the OSI model itself that clearly tells us that layer 2 is the Data Link Layer, which includes devices such as switches, and Layer 3 is the Network layer which would traditionally include routers and bridges.

So, what in damnations, is a Layer 3 switch. By definition this should be called a router, right? That’s certainly what I thought until recently.

In fact there are some subtle distinctions between a Layer 3 switch and a router. These distinctions however make a not-so-subtle difference in performance especially on corporate LANS connected by a VPN – layer 3 switches are blazing fast and have a throughput that would even make Takeru Kobayashi, the Japanese hot dog eating champion, blush. The ability for these devices to “ram” massive amounts of data through them is the main difference between a router and a layer 3 switches.

Hardware implementation of carefully refined software algorithms is what makes this all possible. By hard coding what would normally be a software implementation on a normal router, Layer 3 switches can attain speeds considerably faster than any normal router on the market and deliver data from across the LAN following the principle of Open Shortest Path First. (OSPF) Most layer 3 switches don’t even have CPUs as a router would, but insted use Application Specific Integrated Chips (ASICs) to get rid of the inherent speed penalty that wasted or tied up CPUs can create.

Additionally, Layer 3 switches typically don’t have a WAN port and are intended to be primarily a switching device that includes routing information. In the real world, this means that data centres using VOIP and/or have a large number of users accessing databases or file systems simultaneously will notice that lag or in the case of VOIP – choppiness – is all but removed.

Obviously, I could go into a tremendous amount of detail on how layer 3 switches are implemented and how the ethernet/MAC fram is bridged to layer 3 protocols such as IP, IPX, apple talk etc, but honestly it really isn’t necessary.

As long as you understand the use of these devices, you will be able to implement them just as easily as any router. I hope this brief explanation helps!

Cheers!