• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Should I enable jumbo frames for iSCSI

I've seen charts showing no gain and sometimes a drop in performance from enabling them.

Are you satisfied with your iSCSI performance with them off? If so, leave it off. If not, turn them on and see if your performance improves.
 
Well just updated both my servers in house to ESXi 4 yesterday, before on ESX 3.5 I had jumbo frames enabled. I kinda of forgot to enable it when I put the servers back into production. I'm going to monitor over next week and record some sats and then enable it and see if I get any better results.

I'm always looking to improve performance. Our SAN isn't the best in the world (MD3000i) but it has recently been upgraded to two controller cards and I added a few new hard drives (1 RAID5 set of 7+1HS SAS drives (3 10k, 5 15k) and 1 RAID5 set of 6+1HS SATA drives (7.2k).

Just use the SATA storage for a few dev boxes and stuff that doesnt really get much use. SAS drives have the production stuff.

I need to load balance the traffic over all the ports a little better first-
iscsistats.JPG


It's set on last used path right now I think, hows load balancing now days for path selection?
 
Last edited:
"Meh"

Unless you're REALLY saturating the interconnect to the point that you need the frame overhead back, don't bother.

Chances are the limits will be in the disks more than the transport anyway.
 
For each of my paths, what is best now to use? right now its just on most recently used. all my VMFS (3 of them) have 4 paths they can take. right now one is active and other is active i/o and other two are standby.

Is it worth using Round Robin? should I just set fixed paths?

Also i have a VMW_PSP_FIXED_AP? what is that
 
For jumbo frames to work correctly all nodes must be enabled to use jumbo frames, and usually you will have to set the new MTU on each device. Didn't see you mention this, so I would put my 2cents out there..

MW_PSP_FIXED_AP=Extends the Fixed functionality to active-passive and ALUA mode arrays.

good luck :)
 
I have run with jumbo frames before just forgot to enable it on the esx hosts after the switch to esxi. I'm going to continue to watch the performance and then enable it once i get some good baseline data.

I turned on round robin to help balance the load across all paths last night, a little more tweaking to do on the MD3000i because its an active/passive and the LUN's can only be owned by a single controller.
 
I have started to notice some more dropped frames (not many) since turning on round robin. Don't think I should be too concerned over them.

droppedframes.JPG
 
from my classes all i remember ever hearing is if your gonna use jumbo frames make sure all your hardware is configured to use jumbo frames and have all of your environment (VC and hosts) running it, mixing can cause lots of issues.
 
got that covered, end to end iscsi can be enabled. haven't done it yet because i'm waiting for my baseline data so i have something to compare it to.
 
Well i have seen a few in the past not many, I never kept too much of an eye on it. I have noticed them increasing since enabling round robin.

Legend.jpg

iscsistats0721.JPG
 
Well on the HCL it recommends I use VMW_PSP_MRU so maybe RR isn't the best idea for the MD3000i
 
All arrays on the HCL are supported with RR. Recommending RR is up to the vendor (Dell does with the MD3000i).
 
You must log in or register to reply here.
Back
Top