Unsolved
This post is more than 5 years old
2 Posts
0
2506
SC4020 w/MXL switch high write latency (peaks of 1000+ ms)
I am investigating poor performance on our SC4020 SAN and i've noticed on ESXi it is reporting peaks of super high write latency on active VMs (large file copies, restoring SQL databases, etc). Looking at the compellent side i'm not seeing the issue at all, highest peak is 20ms.
We have the SC4020 connected to our hosts blades in an M1000e chassis via 10gb SFP+ and MXL 10/40 switches. Looking at the interface statistics for the SAN side interfaces i'm seeing high values for throttles on input statistics. This points me to flow control or possibly storm control, but we are currently set up for DCB for that.
Is my problem because the SC4020 does not support DCB and I need to move to regular flow control? I've found plenty of best practice guides for this and many seem to conflict
linsol
1 Message
1
January 22nd, 2018 01:00
did you manage to resolve this issue? we are having same problem , can you please let me know if you find the fix? we are also using MLNX-OS SX1012 with Dell SC4020
DELL-Bob Mi
230 Posts
0
January 30th, 2018 17:00
I would first check to see if you have Delayed Ack enabled in ESXi. Per page 9 of Dell EMC SC Series Best Practices:
iSCSI initiator settings for delayed ACK: During periods of high network congestion in some environments, iSCSI transfer latency may exceed acceptable levels. VMware recommends disabling delayed ACK.
Please download the ESX Best Practices:
Dell EMC SC Series Best Practices with VMware vSphere 5.x–6.x
http://en.community.dell.com/techcenter/extras/m/white_papers/20441056
Per the Compatibility Matrix this model of switch has been tested for compatibility:
http://en.community.dell.com/dell-groups/dtcmedia/m/mediagallery/20438558
I would open a case with Dell Support to get specific answers for your system.
jwilliams-murdochs.com
1 Message
0
April 26th, 2019 10:00
Any update on this? We're experiencing the same issue with our SC4020, and the throttles we see on the SAN switches interfaces are constantly growing.
Did disabling delayed ACK fix the issue for you?
Thanks for posting