high workload on the virtual machine due to the multiple lun connection

[tvbsystem]

Hi

I have  a question about the operation manager showing that the work load is high due to the

one of the lun that is attached is having high IO.

For example

Physical server A  is using LUN 1　and LUN2 and Virtual server 1 is on lun1

Physical server B is also using LUN1 and LUN2 and virtual server 2 is on lun2

And when this virtual server 1 starts using high IO

the Operation manager detects that Physical server A Lun1 workload is high.

But the same time on the Operation manger the Physical server B Lun1 detects that the Server B Lun1 work load is high too do to the Physical Server A virtual server 1  is using.

In this case will  the virtusk server 2 which is on Physical server B  will get some effect by the physical server A virtual server 1 on lun1 is useing high IO?

or this high work load only shows the fact of the indivisual phsyical server and virtual hosts?

doese any one know the details for this showing of the operation mansger?

thankyou

hiroki

[mark_j]

If the luns share the same underlying storage/disk, you have finite disk resources and one VMs on one lun may impact another VM on another LUN. The workload for vms/datastores is configurable to include disk IO in general, as well as what specific attributes to take in consideration. You typically have read/write rates and iops, as well as outstanding IO.. it is possible the outstanding IO is where your workload is being impact by load being placed on the underlying storage hardware, thus impact VMs on LUNs other than the immediate LUN generating the load. Same goes for a physical server sharing the storage subsystem.. load is load, and if OIO or latency is increased/impacted by any additional load (physical or virtual), you'll see the ripple effects in the virtual environment. You can graph the OIOs to see if this is actually fluctuated and thereby the cause of your symptoms/alerts. If you don't care about Disk OIOs, simply deselect that attribute in the workload area of the active policy, and it won't be taken in to consideration for workload anymore.

[tvbsystem]

Hi Mark.J

Thankyou for you relpy.

so the what your saying is that the if the LUN is over lapping on the physical hard drive, if one of the LUN gets high I/O the other will be also be effected.

and for the alerts for the no used LUN the esxi i can deselect and dont worry about it?

Am i getting what your trying to say correctly?

Sorry for may bad English.

hiroki

[mark_j]

If I understand you correctly, it just sounds like you've got shared storage with multiple LUNs, where one LUN's load is impacting the others in terms of latency increasing as the shared storage gets Disk IO. If so, that is a very common scenario and yes I wouldn't be surprised by it.