05-05-2011 12:59 AM
since we are trying to find out if our FC switches have performance issues ( Support by Netapp is telling us that it might be a fabric problem causing high CPU loads on our filer) i installed DCFM pro and tried to see if there is any problems with the switches. Performance graphs are looking fine, nothing is at limit, i dont see errors or problems.
now that im at it, the system has not been updated for quiet a while since it was bought.
After checking out this : http://community.brocade.com/docs/DOC-1642 i was wondering how i get to the downloads to upgrade the OS from 5.1 to a recent release step by step cause i dont see any download links in MyBrocade. Second question is, is it really necessary to touch a running system or where do i see improovements ?
05-05-2011 01:03 AM
If you bought a support contract with Brocade, you can add the serial to your accvount and get access to Firmware.
if you bought the switch from EMC or HP you should go to them.
HP for example has the firmware publically available.
05-05-2011 01:06 AM
Sorry forgot to answer about the second part.
Usually a higher firmware level fixes and improves older levels it also introduces new errors which have not been discoverd yet.
In your case, what did NTAP tell you excatly? What clues or arguments did they hand to you....?
05-05-2011 01:09 AM
I reported high CPU loads there without a higher number of maschines causing it. then i got this:
" When the CPU is at 50-60% there is moderate I/O and OK response times. When CPU drops to 10% and systems is nearly idle, that means that the
amount of requests that receive the filer has gone down."
"That sounds like a "fabric" issue, so anything between and including host and filer target HBA's."
im not sure where to look for errors now, ESX Hosts have 2 FC hbas connected to 2 Brocade E200 which both have a seperate link to the Netapp filer. I dont see any high bandwith or errors.
05-05-2011 01:47 AM
Monitored with Nagios, the filer "suddenly" started to increase CPU usage more often, steadily. Like it was 30 usually with around 120 VMs and then it started to be 50%-60% all day long with little breaks and we were not sure how it happened. Usage on the VMs did not change. Traffic did not change.
Some maschines have slower disk performance, thats what our feeling was. Trying to find the bottleneck now
05-05-2011 01:58 AM
Did you crosscheck the NTAP graphs from Nagios with Graphs from ESX and switches?
Did traffic steadily increase or decrease any spikes, does ESX/VM's show unusual behaviour before or during your time frame?
05-05-2011 02:13 AM
not with switches, but with ESX, i dont see a reason for the heavy load. thats why i thought the switch might have lots of things to do to cause traffic. but it doesnt look that way. we are guessing its internal ntap re-arrangment cause of full aggregrates or so. still checking that...
no unusual behaviour so far on the esx or the vms so far.
by the way: upgrade path for e200 according to that public HP website: 5.1 -> 5.3.0d -> 5.3.1a -> 6.1.2b -> 6.2.2e but im missing 6.0 right ? HP doesnt offer it but if i understand it right i need it to upgrade 5.31 to 6.0 first before i can get 6.1.2 correct ?
05-05-2011 02:29 AM
What did NTAP say other then a Fabric problem, arguments, known problems, conclusions (based on your figures) .
I'll get back to you about the firmware based on HP, but do you have hp switches?
05-05-2011 02:43 AM
you can upgrade directly to 5.3.1a from 5.1.*.
From that point on its a one step upgrade proces..
More information and older firmware can be found on