vSphere Storage Appliance

 View Only
  • 1.  Storage optimization across the Datastore

    Posted Mar 01, 2012 02:15 AM

    Hi,

    I have created two datastore sizing 500 GB each.

    My question that;

    1. In one of the datstore I need to extend the storage say datstore1.Kept two VM each from esxi host in the cluster.

    2. The datastroe 2 ius underutilized.Can datstore1 borrow space from that?If yes how this needs to be done.

    3. If I create antoher datastore3,can I add this storage to datstore1?Do I need to extend volume or something else here.

    What is the best way to achieve that.

    Regards,

    Sushil



  • 2.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 02:34 AM

    1. In one of the datstore I need to extend the storage say datstore1.Kept two VM each from esxi host in the cluster

    Am not sure what you are asking - if you are asking can two VMs be stored in the same VMFS datastore the answer is yes

    2. The datastroe 2 ius underutilized.Can datstore1 borrow space from that?If yes how this needs to be done

    No - the LUN sizes are fixed

    3. If I create antoher datastore3,can I add this storage to datstore1?Do I need to extend volume or something else here.

    Yes you can create an extent that would expand the size of the VMFS datastore bute I would not recommend doing this - preferred method would be to create a new datastore sized approriately and move the VMs to te new larger datastore -



  • 3.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 03:02 AM

    1. The datstore1 is 500GB.The LUN1 size of this is 500GB.Here 2-3 VM are running and it is heavily loaded and surley will exaust this space in future.The datstore2 is mapped with lun2 of 500GB.It has space free in it.My question is to borrow from datstore2 to datstore1.

    3. We have created Lun3 with 1.95TB hence datatore3.It doesn't allow to add more than that.My dtastore is also 1.95TB.This would consists of database.Wondering how to have LUN size more than 2TB in turn single datastore than 2TB.How to achieve that?

    Is it best practice to have one datstoore per LUN?What about vmotion and FT.Does the vm1 within the datstore will be able to move or can it be moved in other datstore as well.How about HA when any ESX machine goes down.

    Regards,

    Sushil



  • 4.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 03:22 AM

    1) Unless you provisioned your virtual disks as thin disks you should not run out of space on the VMFS datastore because by default the virtual disk files (VMDK) are created as the full size so in other words when you create an 8- GB disk an 80 GB vmdk file is created. The case of the thin disk provisioning a vmdk is created and will grow to the full size of the virtual disk - the bigger concern on storage is not saturating the paths to storage or staurating the LUN

    2) Are you running vSphere 4 or vSphere 5 - if it is vSphere 5 the maximum zive of VMFS datastore is now 64 TB but the virtual disk is still capped at the 2 TB -512 B

    Yes best practice is to have a single VMFS datastore per LUN - with vmotion as long as both hosts can see the the datastore vmotion is possible as long as the other vmotion requirements are met.

    Since FT tolerance requires two datastore this is not an issue

    Using storage votion yoou would be able to move the VM between the datastores while it is running

    Like vmotion as long as both ESXi hosts can see the datastore HA will work



  • 5.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 03:49 AM

    1. Thin disk needs to be enabled on storage or while creating VM?You mean 8GB will create 8GB not 80GB if i am not wrong.

    2. I am using esxi 4.1.0.What is the limit on this for datastore.ATnyways very curious to know if LUN5 say is of 1.85TB then how come the datastore of 64 TB in version5.My understanding is that if you have LUN of 1.95TB then same would be the size of datstore if we creating only one datstore.How this works.Little confused on this.



  • 6.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 04:04 AM

    1) Thin disk is set at the time of VM creation and yes I meant 80 GB

    2) To make sure we are using the same terminilgy - LUN is the unit of storage being presented by the RAID/SAN - LUN is then formatted with the VMFS file systems and is used to store virtual disks - the virtual disks also referred to as VMDK files - the limitations of vsphere 4 limits the size of accessible LUNs to 2 TB - 512 B which then can be formatted as a VMFS datastore of 2 TB -512 B you then can carve that up by crfeating VMs with virtual disks of varying sizes.

    With vSphere 5 VMware has changed the access to the LUNs allowing ESXi5 hosts to access LUNs 64 TB in size and format those as 64 TB VMFS datastores -



  • 7.  RE: Storage optimization across the Datastore

    Posted Mar 01, 2012 05:07 AM

    1. why 80GB VMDK while creating disk of 8GB? So Ig I create disk of 100GB then it would be would be really higher then...



  • 8.  RE: Storage optimization across the Datastore
    Best Answer

    Posted Mar 01, 2012 11:05 AM

    Kumar, while creating a virtual disk you have the option of choosing from the multiple disk formats:

    ========excerpt============

    • zeroedthick (default) – Space required for the virtual disk is allocated during creation. Any data remaining

    on the physical device is not erased during creation, but is zeroed out on demand at a later time on first

    write from the virtual machine. The virtual machine does not read stale data from disk.

    • eagerzeroedthick – Space required for the virtual disk is allocated at creation time. In contrast to

    zeroedthick format, the data remaining on the physical device is zeroed out during creation. It might take

    much longer to create disks in this format than to create other types of disks.

    • thin – Thin-provisioned virtual disk. Unlike with the thick format, space required for the virtual disk is not allocated during creation, but is supplied, zeroed out, on demand at a later time.

    ========excerpt============

    Thin-provisioned disks provide you with space efficiency, however you also risk running out of Datastore space if you have overallocated the Datastore. Starting vSphere 4.0, there are alarms which you can use to monitor the space overallocation (not enabled by default) and also the Datastore usage (enabled by default).

    I believe the feature is neat and extremely useful. We have long been using thin disks in our prod environment with no issues. But you should ensure to have a process in place to monitor the overallocation, avoid running defrags in your Guest OS, and etc. I would suggest you to go through the below KB and the included pointers to whitepapers if you intend to use Thin-provisioned virtual disks:

    http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005418

    Best of Luck!!