[c-nsp] Cisco 650x sup2 / sup32 configuration - what makes sense?

John Gill johgill at cisco.com
Fri Dec 9 13:08:51 EST 2011


Hi Jeff,
There are multiple layers of ASICs here, but yes 24 ports do share the 
1G uplink but 8 ports share the buffer.  There are two top level ASICs, 
spread out into 6 lower level ASICs.  The top level ASICs go to the DBUS 
and the lower level ASICs go to the ports and the lower level has the 
buffering.

Did you check to see if you have unicast flooding at the time?

Before deciding what to do, the problem needs to be udnerstood first, I 
am unclear on the root cause of your observation.  It does not sound 
normal at this point.  However, if you do not intend to use this card 
for user access, it is not hard to see you will hit limitations easily 
in a server closet or datacenter.

Regards,
John Gill
cisco

On 12/9/11 5:39 AM, Jeff Meyers wrote:
> Am 08.12.2011 11:18, schrieb Gert Doering:
>
> Hi,
>
>> The best choice? Don't use 6148-GE-TX modules. They are fundamentally
>> broken (8 ports share one ASIC with a single-GE uplink, one port that's
>> "full" will block out the other 7 ports, ...). It's even worse if
>
> I've heard that now multiple times, but a # sh int cap says:
>
> Switch-1#sh int cap mod 5 | i ASIC
> Ports on ASIC: 1-24
> Ports on ASIC: 1-24
> [..]
>
> So does that mean, that even 24 ports share 1x Gigabit? As I said: a
> fully utilized port 14 had a negative side-effect on more or less all
> ports on the blade but certainly on port #4. My understanding is, that
> this should not be the case if the ports 1-8, 9-16 and so on build a group.
>
>
>>> Does it make sense to replace the 6148-GE-TX with a 6748 or is the sup2
>>> respectively the sup32 the actual bottleneck?
>>
>> 67xx won't work with the sup32 (thanks, cisco). I'd go for 6516-GE-TX,
>> which have a much saner architecture than the 6148-GE-TX *and* will
>> work with non-fabric-enabled supervisors.
>
> What about 6148A-GE-TX or 6548-GE-TX? Will one of those modules help me
> to solve this issue or at least make it less worse?
>
> Thanks!
>
> _______________________________________________
> cisco-nsp mailing list cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>


More information about the cisco-nsp mailing list