Identifying the storage is a bottle nack ?

[zeevik]

Hi.

I'm pretty familiar with VMware ESX hosts and vSphare configuration.

Storage was (and still is) the weak side for me.

How can I identify thet the storage is a bottle nack, in terms of performance ?

Are there any statistics on the ESX host side which can tell me that the VMs are slow because the access to the disk is the problem and not CPU/Memory ?

I'm working with local ESX storages, NFS based storage and now starting to work with iSCSI.

Thanks,

Zeevik.

[Kasraeian]

Hello,

In vi/vsphere client go to performance tab for your host, in the switch to box select "disk"

Select "chart options ...", then select the storage you want to monitor from "object" and select the latency parameter from "counters".

Now watch the chart and decide if the latency of reading and writing is high, may be the storage going to be your bottle neck.

-= If you found this note/reply useful, please consider awarding points for "Correct" or "Helpful" =-

-= If there's any mistake in my notes, please correct me! =-

-= Thanks =-

MCTS, VCP

[zeevik]

What should be normal values (mainly, peak ones) for latency on a "good" storage ?

Thanks,

Zeevik.

[Kasraeian]

Its vary, depends on Storage and the environments (system usage).

For example the speed and ability of HP MSA 4400 and EVA 8400 is different, also the utilization of Oracle DB in some Governmental Organization would be different from SQL server in some University or Non-Governmental Organization.

you must check it with your expectation from your softwares and your HW.

If I miss anything please correct me.

-= If you found this note/reply useful, please consider awarding points for "Correct" or "Helpful" =-

-= If there's any mistake in my notes, please correct me! =-

-= Thanks =-

MCTS, VCP

[zeevik]

We're experiencing from time to time a feeling of "frozen" VMs, mainly Win-XP VMs on which people are working with RDT.

I'm trying to identify the root cause for that, and my main suspect is high load on the storage, which leads to low throughput and high latency.

It is very difficult to identify the timing of that,

and it is based mainly on people feeling the VMs are slow.

This is why, after I know how to get the info (thakns again), I want to know what latency is considered high and may lead to the "freeze" feeling

(my guess is latency on accessing the windows virtual-memory files)

Thanks,

Zeevik

[Kasraeian]

Can you please give more info about number of user (concurrent user) and the VM config?

Did you also Monitor the CPU and Memory usage from inside of your VM? (if they use a lot of CPU or Memory, then you should change the VMs config a little so they will get more memory and cpu, or even moving them to the other host)

Did you add any memory in order of testing if the VM performance getting better?

-= If you found this note/reply useful, please consider awarding points for "Correct" or "Helpful" =-

-= If there's any mistake in my notes, please correct me! =-

-= Thanks =-

MCTS, VCP