[c-nsp] 2960-24TT-L or larger?

[Rolf Mendelsohn]

Hi Adam,

I think I can provide some input here...

On Thursday 19 April 2007 21:56, Adam Greene wrote:
> Hi,
>
> I'm planning to deploy a wireless repeater with (4) 90° sectors which will
> backhaul over a 100Mbps full duplex link to my head end, where I have
> 100Mbps to the Internet.

> I think realistically, I can expect the repeater to generate about
> 45-60Mbps traffic over the backhaul.

I assume you mean Wireless base station (Wimax or similar).

> I only need L2 functionality at this repeater, as customers connecting to
> each sector will be split into unique 802.1q VLANs which will tie back to
> the head end L3 device.

If you want to have more scalability and have a Graphable, more easily 
manageable service, the more Vlan's the better.

> I think a WS-C2960-24TT-L will be an adequate switch to tie the sectors and
> the backhaul together at the repeater, but I'd be interested in real-world
> experience as to whether this switch will start to melt at those bandwidth
> levels. Customers will be businesses mainly engaging in typical Internet
> traffic (HTTP / FTP / SMTP / POP3 / VoIP). There may need to be some QoS
> marking / queuing going on at the switch as well.

2960 is good but it only does 255 Vlan's
2400ME does 1000Vlan's, but doesn't allow for VTP (if anything possessed you 
to use it :>(.

3560 does L-3 and 1000Vlans.

> I guess the next step up would be the WS-C3560-24TS-S. Like the
> WS-C2960-24TT-L it's rated at 6.5 Mpps but unlike the 2960 it has twice the
> switching fabric (32 Gbps vs 16 Gbps). Not sure if this really makes any
> difference in this application.

I assume that doing lots of QoS on the switch) where there exists no limit to 
speed (in comparison to Wireless Base Station vs. Slower backhaul) is really 
*meaningless* - i.e. do limits / fancy queuing on the Wireless stuff if it 
supports it (most does).

cheers
/rolf