Morning,
We recently acquired a used md3000i with an md1000 added shelf. The 3000i has 15x450Gb 15k SAS (in one RAID6 LUN) Dell drives. The 1000 has 15x 1Tb 7200rpm (in 2 RAID 6 LUNs).
I have this hooked up to two clusters of ESXi machines using multipathing and iSCSI on a dedicated iSCSI switch. This switch also has our original SAN which is similar but more of a mishmash of drives.
The two clusters are 4xr710 and 4xPE2950.
Performance to the original SAN is ok. It has too many VMs on it and I need to upgrade the firmware. So we got the 2nd SAN so I could update the first SAN without powering down 60 VMs.
I updated the unit to the latest firmware:
Current configuration
Firmware version: 07.35.39.64
NVSRAM version: N1532-735890-005
EMW version: 03.35.G6.50
AMW version: 03.35.G6.50
I have successfully copied data to the unit and it was performing great until about 2 weeks into production when everything shat the bed. VMWare storage migration wouldn't even migrate the machines out because it was timing out. I had to SSH into each host and manually move the machines from the new san to the old san. This process took a really long time. I guessed it was transfering at about 5mb/sec. The average latency on LU0 was over 90seconds (90,000ms), which of course would cause vMotion to time out.
Now I don't know what kind of tests I can perform on this, that's what I'm looking for. Trying to see if it's a controller issue or if it's a drive issue, back plane...I don't know but I would like to find out.
Thanks!
Josh!