on MDS SAN switch, to find out if there are any performance issues, and what are the cause?

[ciscomdsmagic]

We have performance issues to access FC LUNs in a 2-MDS switches environment. On the storage (NetApp), we don't find any latency, or any obvious issues. My questions, on the switch side, what steps can I do to troubleshoot any performance issues, any commands or what aspects should I look through?

Thanks very much! 

[dynamoxxx]

do you have DCNM and is it configured to collect performance stats ?  Is every system connected to this NetApp array having issues ?  Do you have other arrays that have acceptable response times ?

[ciscomdsmagic]

I don't have DCNM. I am seeking for native MDS tools or commands or methodology to troubleshoot on the switch, as I asked early. I only have NetApp array.

Can anybody else please help?

[dynamoxxx]

care to elaborate on other questions i asked ?  That is if you want help.

[ciscomdsmagic]

this is a new implementation, so, we have not widely put too many hosts yet.

can anybody else please help? I really just need help on any native methods or commands? not anything fancy.

[dynamoxxx]

but the ones you did implement, are they all having issues ?

[Bill Wilson]

If you have latency for the host attached and latency from NetApp then you know how long its spending on the fabric.

Fire up Device Manager for both switches, check the log for any errors or discards showing for the FC ports being used. This is mainly looking for any SFP or cable errors.

Also check the buffer credits, if you're running out of them it will impact performance.

Sorry its not much but without further info its difficult to comment.

(I'm assuming that the host shows no errors on the HBA)

[ciscomdsmagic]

Thanks Bill for the message. This is something I am looking for.

we don't see errors about SFP or cables. What about the length of the queue on the port? How do I check if it is too much waiting on services?

Also, how do I check the buffer credits, and if it running out or not?

[Bill Wilson]

Command to see current bbcredit for all FC interfaces is

show interface bbcredit

You can specify the interface for the ports you want to look at.

The following is a good overview for port "problems"

http://www.cisco.com/en/US/prod/collateral/ps4159/ps6409/ps12970/white_paper_c11-729444_ps4358_Products_White_Paper.html#wp9000261

If you can't find anything on the switch I'd take a loook at your host HBA setup, especially what its queue lengths are set to.