Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1313
  • Last Modified:

Storage Performance MD3000i + vSphere 4.1

Hello,

I run two Clusters between two different Datacenters and networks. One of them (the fast one) is running ESXi hosts with 4.1 version and the other one (slow) is running ESXi on 4.0 version.

One of the clusters I have lots of more VMs than the other and customers are not complaining of storage performance. On the other cluster I have much less VMs and customers are complaining because of storage performance.

When I start to deploy a template at the same time between both Clusters, I see that the cluster without performance issues goes much much faster than the other.

The fast cluster is using 15x450GB SAS disks in one group only of RAID 5 plus one hot spare. The other (slow) cluster is using 15x300GB SAS disks one one group only of RAID 5 plus one hot spare. I know that one group only is not the best way to have performance after reading some MD3000i documents, although I cannot do anything at this time. The weird thing is that one of the cluster is running without any issue and the other is having performance issues.

How can I check on the new vSphere version my storage performance ? Maybe one of the VMs is killing all the storage I/O ? How can I check that ? I couldnt find it on VMWare manual.

If its not any VMs can it be any faulty disk or a slower disk in the storage degrading all the performance ? I dont remember if I have any 10K disks in the storage. Is it possible to check if any of the disk on the array is degrading performance ? If so, how ?
0
maxihost
Asked:
maxihost
  • 7
  • 4
7 Solutions
 
jesse_7271Commented:
In VI Client

Click on the esxi host, click on the performance tab, choose advanced from upper left, then choose disk from the Switch To:  combo box

Test3.JPG
0
 
jesse_7271Commented:
If you are looking at host usage you could also use Administrative Tools -> Performance Monitor to check if one of your VM's Avg Disk Queue is pinned.
0
 
jesse_7271Commented:
You could also use the normal overview, then select Virtual Machines to view the Disk KBps used by each VM.
0
Concerto Cloud for Software Providers & ISVs

Can Concerto Cloud Services help you focus on evolving your application offerings, while delivering the best cloud experience to your customers? From DevOps to revenue models and customer support, the answer is yes!

Learn how Concerto can help you.

 
maxihostAuthor Commented:
Hello,

Thanks for the help, but I could not find how can I check for which VM is affecting the whole storage.

For a KBps usage on each Virtual Machine, what is too much on an average for the last day ? Is it possible to measure like this ?

Guilherme
0
 
jesse_7271Commented:
That is a complicated question which is dependent on how many disks/what kind of disks/what kind of IO Sequential, random, read, write, what RAID Level, what speed is your fabric, etc.  That gets kind of sticky once you get into it.

If it was me, I would actually use performance monitor inside the VM, then check for the avg disk queue(green line by default), and see if one of your VM's is pinned.

Also, snapshots can sometimes cause issues, are you sure you don't have any open snaps?
0
 
maxihostAuthor Commented:
Jesse,

I think I found the issue. One VM is consuming LOT more than the others.

Look the file attached.

I am moving that VM to a local disk to see if the issue has been solved.

Guilherme
untitled.JPG
0
 
jesse_7271Commented:
Nice,

Looks like it kicks into over drive at about 4:00am

You might want to change your time period to 2 weeks, just two see if there is a pattern of it getting worse or remaining static.
0
 
maxihostAuthor Commented:
Jesse,

I have moved that VM to a local disk, although the storage is still not satisfatory.

Any other ideas ? There are no VMs anymore consuming that much.

Guilherme
0
 
jesse_7271Commented:
As in, that VM or the entire rest of the area?  Have you tried playing

Are you sure the disks are SAS(not SATA)?
Are you sure of their speed?
What is your connection method, Direct attached?iSCSI?
Do you have a login or management interface on the MD3000i?

Does it display the disk latency(in ms) on the VM->Performance tab?
0
 
maxihostAuthor Commented:
Jesse,

I have moved only one VM to the local disks. It seems that is a little faster (the storage), but I didnt understand yet how to measure storage performance, I know the latest version of vSphere have some kind of storage control, but I dont know how to use it. I have enabled on one of the LUNs (storage control) and its set to 30ms, what does it mean ?

- The disks are SAS, I am sure. I am not sure if some are 10K and some 15K or if all them are 15K
- Its iSCSI
- I have login / MDSM to MD3000i

The average latency (in ms) of all the hosts is 20ms.

Guilherme
0
 
jesse_7271Commented:
Check some of these reports
disk1.JPG
disk2.jpg
disk3.JPG
0

Featured Post

Concerto Cloud for Software Providers & ISVs

Can Concerto Cloud Services help you focus on evolving your application offerings, while delivering the best cloud experience to your customers? From DevOps to revenue models and customer support, the answer is yes!

Learn how Concerto can help you.

  • 7
  • 4
Tackle projects and never again get stuck behind a technical roadblock.
Join Now