Which Switch to get 3560vs3750vs2960 or better?

[Justin Bray]

We are currently running two stacked 2960-s switches that is conecting our SAN to our netgear access switches. We our currently looking into either replacing the 2960s with 3560 or 3750. Has anyone delt with these models before and can give me a realworld overview to help us make our decision. We would also like to be prepared for future deployment of VoIP.

[Joseph W. Doherty · Accepted solution]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

The 2960/3560/3750 are all known for often having insufficient buffer resources when QoS is activated.  If you're looking use QoS on congested ports (SAN?), you might want to look at a 4900, 4500 or 6800 series.

PS:

I work in a production environment were we have almost any L2/L3 Cisco switch in the last 15 years.  I work with 2960s, 2960-Ss, 3560(/G/E/X)s and 3750(/G/E/X)s.  They're usually are fine for access switches under typical access light loading.

[Leo Laohoo]

I'm going to agree with Joseph.  2960, 3560, 3750 & 3850 can't withstand the continuous data traffic between SAN array.  It's OK to use either one of them for iLO.

Aside from 4900, 4500, 6500 or 6800, you also need to weigh in the "size" of your SAN.  If it's big, then you need to go up to the Nexus solution.

[Joseph W. Doherty · Accepted solution]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Since my original posting, I had to deal with a 3750X dropping packets like crazy with QoS enabled and hosting SAN devices.

I was able to tune the buffer settings to obtain performance even better than with QoS disabled; got the drop rate almost to zero.

My approach was to not reserve buffers (i.e. let the common pool have them), set thresholds to maximum (maximum increased on later IOS versions), and increase buffer allocation to the queue suffering drops (in my instance, the default class).

Basically, one port queue could then take all the buffer space on the ASIC.

In my case, my problem was transient congestion on just a couple of ports, not a huge amount of contention between ports.

Also on the 3750X, there's 2 MB of buffer space per 24 edge ports or per 4 uplink ports, so host placement can be important too.