Hello,
we have two IBM 5045 enclosure's with each 12x 3,49 TiB SSDs. 3x VMWare 7.0.3 hosts are connected with 10GbE ethernet iscsi nic's (dac-cable).
Firmware 8.7.0.1
We configured this hosts as generic-hosts without chap. We disabled DelayedAck on the esxi hosts. No jumbo frames.
If we copy big files form one VM to a second VM we see 400-500MB/sec.

IOMeter and diskspd show 35.000 iops

But our backups a terrible slow 10-15mb/sec. The storage latency is very high. esxtop show DAVG 184824ms. You can't control the while the backup is runing.

DAVG => This is the average response time in milliseconds per command being sent to the device.
KAVG => This is the amount of time the command spends in the VMkernel.
If we stop the backups the system is working normally.
Today we stressed the storage and copy a lot of gigabyte from internal storage to ibm and backward. And we se the same issue. VM hang, big davg / kavg vaules and after a some time it works normal.
We see 100% active time for the hard disk but no read or write:

The IBM system works in hyperswap mode. The problem exist with single disks too.
It seeams that the san traffic got a stop signal from the storage
Please help.
Thomas
------------------------------
Thomas
------------------------------