EtherChannels on a Cat4000 - how many & where?

I assume you have spanning-tree running ?

If so it will block VLANS on one of the Etherchannels and as a result you could end up putting traffic over to the new Etherchannel immediately.

The best way to do this is during a schedule network maintenance window.

Actually, I believe that we are *not* running STP as it causes conflicts with AppleTalk, and we have a lot of legacy AppleTalk stuff to support for another couple of months.

I've taken your suggestion and will try scheduling some maintenance for tomorrow morning. I'll make sure neither switch has a default route, then disconnect the first EtherChannel and bring the other up; then try and ping 'across' the switch.

Chris

Does that switch run IOS or CATOS - I cannot remember

CatOS...

anyways for future if you were doing this on a L3 switch then you could turn up the new ports as routed interfaces with non-routable IP address

- non-routable meaning given an address that lies outisde of the network statments under the routing process being used.

Then you can pre-test the links by doing long-running pings to ensure the link is problem free. You can do this in a L2 environment but it take much more care...

Because it's an old (and crappy) switch, it only has a L2 Supervisor engine installed. I'm actually curious as to what the benefits are of having an L3 Supervisor engine are?

We essentially have one big, busy local network and then not-busy remote networks that connect to us (via cable, T1, etc). We need to maximize throughput between the local machines, the remotes aren't as important.

Chris

One of the key addvantages of a L3 switch is that it allows for the segmentation of a network ( read different IP subnets). Flat networks are typically problematic for a variety of reasons.

Network segmentation is particularly important for wide area network link that are relatively slow speed so one does not want unnecessary traffic (braodcast/multicast, etc) traversing those links.

You mean, I can divide up my network into smaller nets and, with an L3 engine, essentially have a 48-port router? What's the difference between what I can accomplish with a L2 engine using VLANs and an L3 engine?

Thanks!

Chris

A L2 switch cannot route traffic between VLANs - a router of some form is required to perform this function. When Ethernet switches were first introduced this was performed by an external router. This setup was oftern referred to as a router-on-a-stick as quite often there was one phsical link between the router and the switch

Later on, on larger Cisco Catalyst switches (ie. Catalyst 5000) integrated routing modules (RSM) wee introduced along with the something called multi-layer switching. The RSM routed the first packet in a flow, populated a hardware table used for swithcing and all packets in the flow after htat did no get forwarded to the RSM.

Integrated L2/L3 switches are the latest iteration and as the name implies they can perform L2 switching or L3 routing based on how they are configured.

So on a l2/L3 switch one can configure multiple VLANs as appropriate and also have the switch route inter-VLAN traffic.

And yes you could have a 48-port router if each switch port was configured as a routed interface.

Okay, that makes sense -- thanks. I hadn't heard of "router on a stick" before... :)

I am lead to believe that the 4000 and 5000 series are pretty similar, and I'm pretty sure that I can throw in an L3 RSM into the 4000 chassis. I need to verify that though...

You've been a great help in this thread, Merv... would I be able to email you offline with some specific questions? Or would you prefer to keep the discussion here?

Chris

I will send you my private email id as I don't read the one that I post with here

Thanks!