<![endif]><![if gte mso 9]>
<!--
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{mso-style-parent:"";
margin:0in;
margin-bottom:.0001pt;
mso-pagination:widow-orphan;
font-size:12.0pt;
font-family:"Times New Roman";
mso-fareast-font-family:"Times New Roman";}
p.MsoPlainText, li.MsoPlainText, div.MsoPlainText
@page Section1
div.Section1
-->
I have
been noticing performance problems on larger LUNs. Basically, most of my LUNs are 1TB with a couple 2TB LUNs. The 2TB LUNs are
LUSEd from 2 1GB LUNs then present to the ESX cluster.
Due to the need for 2TB partitions, I would format
them as an 4MB block storage so that I can present a 2TB partition to the
VM.
When I run tests on smaller (15GB) partitions coming from the 1TB LUNs vs. the 2TB partition coming from the 2TB LUNs, the performance is significantly lower on the latter.
Do you know if this is a problem
from a performance perspective?
Thanks,
Daniel
Daniel,
I can't find it at the moment, but there is a doc from vmware detailing that LUSE LUNs are a last choice. Also, everytime I've ever called VMW support and told them I'm on LUSE LUNs, I get a nice little 'lecture' about how bad they are.
We also see performance 'issues' on them. They aren't slow, per se, because the ridiculous amounts of cache in our USP1100's buffer alot of that out, but they should be faster.
We had no other options beside LUSE LUNs, so I went with that. We are moving away from Hitachi for this stuff, and so I wont have to deal with them anymore.
Feel free to send me a private message if you want to talk in more detail
--Matt
Are the LUNs on the same storage? There should be no difference in performance between VMFS LUN sizes, so I would lay the blame on the storage if I had to guess at the cause.
Hitachi LUSE LUNs are basically the worst performing combination possible. Stick with non LUSE luns...
--Matt
Hi Matt,
It's interesting that you mentioned Hitachi. We are using them here for our SAN. Do you have any reviews or doc on the LUSE performance vs. non-LUSE performance? I want to have something solid to discuss with our SAN team.
Thanks,
Daniel
Daniel,
I can't find it at the moment, but there is a doc from vmware detailing that LUSE LUNs are a last choice. Also, everytime I've ever called VMW support and told them I'm on LUSE LUNs, I get a nice little 'lecture' about how bad they are.
We also see performance 'issues' on them. They aren't slow, per se, because the ridiculous amounts of cache in our USP1100's buffer alot of that out, but they should be faster.
We had no other options beside LUSE LUNs, so I went with that. We are moving away from Hitachi for this stuff, and so I wont have to deal with them anymore.
Feel free to send me a private message if you want to talk in more detail
--Matt