At 4:46 PM +0000 2/25/03, Robert Edmonds wrote: >Layer 3 switching combines the best of switching and routing in one >platform.
It isn't the best of both worlds, because raw forwarding speed is frequently not the constraint on real-world network design. From a cost standpoint, it would be completely insane to put high-performance devices of this type into branch offices. Fast is only one component of a design solution. Before people start worrying about being wire speed or not, look at the requirements. If you choose to put in Gig E simply because your servers can, hypothetically, service 200 Mbps of traffic and you don't want to use Etherchannel, it's irrelevant if you can forward at Gig E speed -- it won't buy you anything. Server interface bandwidth, in my experience, is far more often to be the limiting factor than delay in the forwarding elements, be they routers, switches, layer 2 routers, layer 3 switches, or what have you. Tailor the platform for the job. For example, the 7200 has a DS-3 interface, but (and I'd have to check on the latest NPE specs), historically it can't fill that interface. With typical US pricing, however, the breakeven point between link cost alone (i.e., not considering additional router interfaces) for multiple DS-1 versus fractional use of a DS-3 is around 6-7 DS-1's. That the DS-3 could carry 28 DS-1's if the router could do "wire speed" is irrelevant to the problem. Having the ability to PHYSICALLY INTERFACE to a high-speed facility may be a much more important cost factor than having that interface run at "wire speed." There are other approaches to network speedup rather than accelerating the forwarding rate. I've frequently improved a network by providing a separate L2 switched LAN for backup or synchronization among colocated servers, putting an extra NIC into these servers. We did this before we knew to call that a Storage Area Network. ;-) >The main advantage here is speed. The way it works is, in a >switch you have some kind of layer 3 routing engine (aka route processor, or >RP). For example, the MSFC2 (Multilayer Switch Feature Card 2) is one of >the options available for the Cisco 6500 (and a couple of others, I think) >switches. When the switch receives a packet bound for a different VLAN, it >sends it to the RP. The RP makes the routing decision and puts an entry in >the route cache for the switch. The first packet in a flow is routed and >the rest are switched at wire speed, hence the increase in speed. That's >kind of a simplified view, but I think it gets the general idea across. So, >layer 3 switching is both routing and switching, but faster (usually, >anyway). > >""DeVoe, Charles (PKI)"" wrote in message >news:[EMAIL PROTECTED] >> I am under the impression that switching is a layer 2 function and that >> routing is a layer 3 function. I have seen several discussions talking > > about layer 3 switching. Could someone explain this to me? Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=63764&t=63728 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
