[c-nsp] Cisco 3560X performance in the wild

[Geoffrey Pendery]

We've seen the output drops on uplink, two Gig attached hosts on the
3560X trying to push hard out the 1 Gig uplink.
With out-of-the-box no tuning or QoS, we'd see it struggle to hit 400 Mbps.
Having tuned it to dump all the buffers into one queue for the uplink,
we can hit 800 Mbps now.
This is also a function of how few and how bursty the hosts are, so YMMV.


-Geoff


On Wed, Nov 23, 2011 at 8:30 AM, Dave <dcostell-cisconsp at torzo.com> wrote:
> On 11/22/2011 08:47 PM, Jeff Kell wrote:
>>
>> On 11/22/2011 5:15 PM, Peter Rathlev wrote:
>>>
>>> On Tue, 2011-11-22 at 12:59 -0700, Dave wrote:
>>>>
>>>> I was wondering if anyone has used the 3560X-48T switches and would be
>>>> kind enough to give me the good/bad/ugly on them ?
>>>
>>> We have a couple of WS-C3560X-48T-Ls in use. They seem to function just
>>> as "well" as their 24 port cousins. Any specific things on your mind?
>>
>> We have some C3560X-24Ts (with IP Services upgrades) in production as CE
>> routers.  They work as "well" as the 3560s we had been using.  We don't
>> have any under serious load, no QoS, and no 10G deployments, so no
>> horror stories yet.  Some output drops on some uplinks but doesn't
>> "feel" as shallow as 3560 non-Xs.
>>
>> They run 3560E images.  No idea how much hardware they have in common.
>>
>>
>> Jeff
>
> Jeff,
>
> The output drops were one of the things I was concerned about. How far below
> line rate were you seeing them ? We picked up some 2960s awhile back and
> noticed that we started seeing drops at around 400mbps.
>
> Does anyone know how many ports/asic on these switches and/or what the
> available bandwidth to the backplane is per asic ?
>
> Thanks
>
> Dave
> _______________________________________________
> 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/
>