EDIT:- Seems to be a problem with the Ubuntu server VM, i ran ATTO on the Windows Server and copied directly to this with Cachecade on and it seems to be working as you would expect. Problem appears to be network related and not storage related in relation to the Ubuntu server VM. It just so happened i was always using this as a destination to copy too. I'll update later should i find an answer.
Hi All,
Im having a storage issue on my Esxi 6 Homelab, basically write speeds are all over the place and I have I/O Latency issues, confirmed with esxtop. vCpu's have been trimmed on VM's to make sure its not vCPU ready time thats the cause.
My config is as follows:-
Lenovo TS140 with LSI 9266-8i (Latest Firmware).
RAID 10 with 4x WD RED Hard drives
Cachecade configured with 2 x OCZ Agility 3 (AGT3-25SAT3-120G - On compatability list) in Raid 1 and linked to the RAID 10 Virtual Drive.
Write Back is on (With BBU backup)
Disk Cache is OFF
In the test scenario i have 2 VM's, one a Windows 2012 Server and the other an Ubuntu 15.04 server.
I copy a 300 Gig file to each at the same time over the network to a share drive on each from a real Windows Client.
The surprising thing here is that i get WORSE performance using Cachecade than i do without. If i diasable cachecade i still get reasonably high latency ~400ms, with Cachecade latency is in the 1000's.
Additionally, when using WITHOUT cachecade i get a nice straight consistant copy speed line in the copying box (about 50MBps total for both, 25MBps each) and both seem to work at the same bandwidth rate even though it seems a little slow.
With Cachecade its all over the place, 1MBps then ~85, then down to 50, 20 etc. The copy graph is like a rollercoaster.
So, is it my SSD's not working properly (SMART Passed OK)? Is the RAID10 slow?
Any ideas?
