Thin provisioning / "right-size" database disk

Similar Questions

• Thin Provisioning LUNS and virtual disks

  Hi all

  I recently came across a problem with our VMware View environment, where in vCenter about 20 of our virtual office message that there is no more space left. When I looked at the data warehouses that store with my clones related, they have both said more than 200 GB of free space.

  In an attempt to clarify the environment, but do not go off-topic, some of our pools view are configured to give each virtual desktop a disposable disc service end of 4 GB (Temp and swap files). I noticed that in vCenter, the question asked to VMware for more memory space pointed to these disposable disks.

  I was with the support of a few hours... on the side of EMC Clarion, LUN (which are thin provisioned) also showed full... but VMware support was able to create a 1 GB file on the data store that was full according to the Clarion.

  We sent EMC our papers and they were back and just said the table is fine, the LUN was out of space... but this does not explain to me why vSphere showed the LUN as having 200 GB free. EMC said they did not know and I need to talk with VMware. I'm confused, if the LUN was not always full, how we were able to create the 1 GB file on the data store?

  Curiously, all the desktop computers that do not have the disk of 4 GB disposable were very well... they had no problem. I'm almost thinking, perhaps disposable disc was full of temp and file exchange and that where this problem started, but I can't seem to be able to confirm that the 4 GB.

  Another thing is, began to be blocked on this data store, even if these virtual desktops, have been removed from the files. I'm unable to delete records in the data store. I know that I can restart the hosts to unlock, but let's just delete this store of data/lun individual completely.

  I was wondering if anyone has had similar problems or has any advice?

  Mike

  Teas because there are 2 parts of the area of thin provisioning.

  When you create a thin LUN on the side of the table, regardless of the size, it only * affects * about 1 GB (for Clariion).  Like many parts of this LUN if written, it allocates more.

  Now when you create an om VMFS file, unless you use eagerzeroedthick (which you're not), there still, not all these data in a 1 GB (10GB or other) is actually hit or written to the creation time.  Thus, the table side only sees (for example) hit 16 MB.  Now to delete some files on the side of VMFS.  The table does not know that those "affected" blocks are now free (because VMware doesn't have the ability to communicate that back to the table), so it independently as always in use.  As a result, you end up with a shift of how is used according to the table, and the file system.

  So now consider a 500 GB LUN, who think that the painting is "full", but said VMFS has 200 GB of free.  That 200 GB was probably written at a time in its history, then deleted.  Thus, full to the table, but not the file system.  When you create small files, VMFS is overwhelming the space previously used, which the Board considers acceptable, but as soon as you try to use the space that is not be 'recycled', it barfs, because its out of space on the backend.

• Thin Provisioning not reduce size of backup

  I use ghettoVCB to save some of my virtual machines. I changed some of these virtual machines to use a provisioning in an attempt to reduce the size of backup. When I browse my store of data using vSphere Client, I see that the thin provisioned disks take less space. However, when I go back to a NFS store on a Windows Server 2008 R2, backup sizes are the same as they were when I used thick provisioning. Is this normal on a Windows NFS share? I did specify a provisioning for ghettoVCB.

  I ran sdelete and VMWare Converter to try to reduce the disks; This decreases the size of the disk that I see in the browser to store data, but the size of the backups on the NFS share was always the same.

  I don't have access to an NFS server ATM, but tests on a ZFS volume w/de-dupe that is exported outside like NFS, I can confirm that on the host ESX virtual machine appears as thin provisioned and when you do ls - lha on the VMDK is 8 GB and not 0 GB, but when you look at the free data store no space has been consumed during the creation of this new virtual machine. It all depends on your server and it is the configuration of NFS, I agree with RParker, you'll want to enable deduplication on your NFS server if it is supported. Put into operation end is handled differently on the NFS Server vs on VMFS volume and it dictated by the server NFS itself on how it is configured to manage a provisioning.

  =========================================================================

  William Lam

  VMware vExpert 2009

  Scripts for VMware ESX/ESXi and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

  Twitter: @lamw

  repository scripts vGhetto

  Introduction to the vMA (tips/tricks)

  Getting started with vSphere SDK for Perl

  VMware Code Central - Scripts/code samples for developers and administrators

  VMware developer community

  If you find this information useful, please give points to "correct" or "useful".

• Backup errors with backup of thin provisioned disks on ESXi 5.1

  Hello

  I try to set up a backup solution for two servers in ESXi 5.1. (free version).

  I have installed and configured the ghettoVCB successfully and everything seems to be ok.

  However, when I start to create a backup, the process would end with an error :-(

  It didn't matter if the server is enabled or disabled, I get the same result.

  All data warehouses are local, 2 x free WD500GB with plenty of space.

  There is no snapshot, anything that could (in my eyes) prevent a good backup.

  The server works perfectly, when turned on.

  I'd appreciate any useful comment. Thanks in advance! The log is attached.

  / vmfs/volumes/519fceb8-3d39150e-2D0F-001999eeb339/ghettoVCB-Master #./ghettoVCB.sh f vms_to_backup
  Recording output to ' / tmp/ghettoVCB-2013-05-25_20-56-28-324192.log '...
  2013-05-25 20:56:29 - info: = ghettoVCB NEWSPAPER BEGIN =.

  2013-05-25 20:56:29 - info: CONFIG - VERSION = 2013_01_11_0
  2013-05-25 20:56:29 - info: CONFIG - GHETTOVCB_PID = 324192
  2013-05-25 20:56:29 - info: CONFIG - VM_BACKUP_VOLUME/vmfs/volumes/backup/ghettoBackups =
  2013-05-25 20:56:29 - info: CONFIG - VM_BACKUP_ROTATION_COUNT = 3
  2013-05-25 20:56:29 - info: CONFIG - VM_BACKUP_DIR_NAMING_CONVENTION = 2013-05-25_20-56-28
  2013-05-25 20:56:29 - info: CONFIG - DISK_BACKUP_FORMAT = thin
  2013-05-25 20:56:29 - info: CONFIG - POWER_VM_DOWN_BEFORE_BACKUP = 0
  2013-05-25 20:56:29 - info: CONFIG - ENABLE_HARD_POWER_OFF = 0
  2013-05-25 20:56:29 - info: CONFIG - ITER_TO_WAIT_SHUTDOWN = 3
  2013-05-25 20:56:29 - info: CONFIG - POWER_DOWN_TIMEOUT = 5
  2013-05-25 20:56:29 - info: CONFIG - SNAPSHOT_TIMEOUT = 15
  2013-05-25 20:56:29 - info: CONFIG - LOG_LEVEL = info
  2013-05-25 20:56:29 - info: CONFIG - BACKUP_LOG_OUTPUT = /tmp/ghettoVCB-2013-05-25_20-56-28-324192.log
  2013-05-25 20:56:29 - info: CONFIG - ENABLE_COMPRESSION = 0
  2013-05-25 20:56:29 - info: CONFIG - VM_SNAPSHOT_MEMORY = 0
  2013-05-25 20:56:29 - info: CONFIG - VM_SNAPSHOT_QUIESCE = 0
  2013-05-25 20:56:29 - info: CONFIG - ALLOW_VMS_WITH_SNAPSHOTS_TO_BE_BACKEDUP = 0
  2013-05-25 20:56:29 - info: CONFIG - VMDK_FILES_TO_BACKUP = all
  2013-05-25 20:56:29 - info: CONFIG - VM_SHUTDOWN_ORDER =
  2013-05-25 20:56:29 - info: CONFIG - VM_STARTUP_ORDER =
  2013-05-25 20:56:29 - info: CONFIG - EMAIL_LOG = 0
  2013-05-25 20:56:29 - info:
  2013-05-25 20:56:35 - info: start the backup for server W2003
  Destination disk format: thin provisioned VMFS
  Cloning disk ' / vmfs/volumes/519fceb8-3d39150e-2d0f-001999eeb339/w2003-server (2) 2/w2003 Server _1.vmdk'...
  Clone: 36% of fact.
  2013-05-25 20:59:34 - info: ERROR: error in the backup of "/ vmfs/volumes/519fceb8-3d39150e-2d0f-001999eeb339/w2003-server (2) 2/w2003 Server _1.vmdk ' to W2003 Server
  Destination disk format: thin provisioned VMFS
  Cloning disk ' / vmfs/volumes/WD500GB-1/W2003-Doma¤ne/w2003-server (2) hard '...
  Clone: 66% of fact.
  2013-05-25 21:08:59 - info: ERROR: error in the backup of "/ vmfs/volumes/WD500GB-1/W2003-Doma¤ne/w2003-server (2) hard" to W2003 Server
  2013-05-25 21:08:59 - Info: backup time: 12,40 Minutes
  2013-05-25 21:08:59 - info: ERROR: impossible to W2003 server from backup because of the error in VMDK backup!

  2013-05-25 21:09 - info: # final status: ERROR: No. VMs saved! ######

  2013-05-25 21:09 - info: = ghettoVCB JOURNAL END =.

  / vmfs/volumes/519fceb8-3d39150e-2D0F-001999eeb339/ghettoVCB-Master #.

  > However, when I start to create a backup, the process would end with an error :-(

  This is not correct.  Depending on your position, you've got to 66%, until he ended with an error.

  I have the same problem. In my case, I make a virtual disk full and mount it with Server 2003 for project a NFS volume.  I use GhettoVCB, as you do.  It uses the standard vkmfstools in the script to make the clone/backup.  I have proven that the problem also occurs when you use the command line vkmfstools.  GhettoVCB has the bad habit of you cache the error code.  The error code is he ran out of disk space, which we both did not happen.  Once the problem occurs, the only way I could fix is to delete the virtual disk and start over, however, the problem will come back.  The problem didn't surface until what I moved to ESXi 4.0 5.1 on the servers.  Through this process, the hardware off the VMS version did not change, nor dare, because I may need to take 5.1 and back to 4.x.

• Question about thin provisioned disks

  I did some searching but can't find specific answers, I need and I apologize in advance if these questions are public knowledge. I know also my setup is very small compared to most houses VMWare normally using these forums, but we are moving slowly in the 2nd Decade of the 21st century (or try).

  I have a new install ESXi 5 and move during one ESX3 host that has about a dozen of VMs on it.  I recently built vCenter host servers and my future view connection server.  I'm kind to vCenter again as my previous installation of ESX 3 was a single server and us were not investing in the vCenter at the time option (missed a lot of features and ease of upgrade during this transition, but 20/20 declining).  I'm a little green in using vCenter so take pitty on my ignorance.

  It's more time than I wanted to and I just want to know the best way to change these 2 servers (recently built and configured systems Win 2008 R2) to thin provisioned disks.  I need to keep all the storage space I can and beyond the obvious problem of being able to provide more space than you physically, I have no problem to do so (in addition to get them changed).  If there are negatives I'm missing, please report them to me.  The possiblities that I see are:

  The clone and thin provision then delete the originals.

  Use the stand-alone converter on them (from above).

  Completely rebuild thin provisioned with disks.

  If someone can give me a quick list of possible problems and pitfalls for these opportunities or to get to a better suggestion, please do so and thanks in advance.

  Manual of Clark

  PRIMOS Inc.

  Since you have storage vMotion the clone or standalone VMware converter are both viable options.   Not sure about the disadvantages of these options.

  You didn't say what kind of server it is, but I'm always careful about the cloning of the domain controllers.   Without the option for storage vmotion I rebuild a DC from scratch.

  Post edited by: mittim12

• Is it possible to migrate a virtual machine to thin provisioned disk?

  I have about 100 VM in a resource pool dedicated to called "Archive"... Well, archiving.  These virtual machines are currently powered off but every now and then we will get pulled for a developer to work on (they are for the most part, the virtual computer with client customizations if they are changed about once a year).  I'm quickly running out of space on this SAN but all discs are thick provisioned and using the disc is maybe 20-30% on average.  Also, I want to move them to a new large store of data I created for them.

  Of course, I choose Migrate & gt; Change store data & gt; Thin provision in vSphere Format and I'm done.  However, it would be nice to not do this 100 times.

  Move-VM does not seem to be capable of this, and I have not been able to find a workaround.  Is there a way to convert the thin and then perhaps Move-VM? Or I'll just on this all wrong.

  Thank you!

  You can try this script?

  He's moving vms to data store 'EMCSAN:Datastore01' trying to change the format of storage for Thin

  $dsView = Get-Datastore -Name "EMCSAN:Datastore01" | Get-View -Property Name
  
  Get-VM  | % {
  $vmView = $_ | Get-View -Property Name
  
  $spec = New-Object VMware.Vim.VirtualMachineRelocateSpec
  $spec.datastore =  $dsView.MoRef
  $spec.transform = "sparse"
  
  $vmView.RelocateVM($spec, $null)
  }
  

  / Yasen

• ESXi 4 Thin Provisioning

  Hello, we have before us to enter the world of virtualization, and am currently evaluating 4 ESXi and vSphere (on the right, the only host for the moment with local 400 GB SAS (RAID10) drives).

  I have a computer configured to drive Thin Provisioning virtual. The hard disk settings show that a currently stocked 10 GB size:

  

  Yet, the summary of the resource for this machine shows 11FR as being currently configured:

  And, more annoyingly / important that the disparity that above, if I try to copy a file of 1.8 GB on this machine, it runs out of space. This raises two questions:

  (1) why it would run out of space when 11 - 7.82 = 3.18 GB is theoretically available?

  (2) even if it did run on space should not commissioning it automatically allocate more space, so this does not happen? As you can see there is lots of space available in the data store.

  I would appreciate a lot of the answers / advice you can offer as despite many VMWare reading guides I can't fothom this one out.

  Thank you

  Sqwim

  What is the disk in Windows within the virtual machine? How freespace do you have here?

  Daniel

  ---

  If you find this or any other answer useful please consider awarding points marking the answer correct or useful

• ESXI 5.1 "the disc is not thin-provisioned" after copying to the new data store vmdk

  I wanted to create an external backup as a 2nd VM ready-to-run on the 2nd data store.

  I've removed all snapshots, stop the machine virtual and exported the OVF file to a computer, adjusted to the size of the original virtual machine starts in Gparted to update the partition and restart. All very well.

  I then tried to deploy the OVF in the data store alternative (235GB slim, thickness according to the deployment of 250 Wizard). I tried both thick and thin but the error message "cannot deploy the OVF. The operation was cancelled by the user.

  I then manually downloaded the VMDK & OVF file to the 2nd data store and tried to inflate but then received the message "a specified parameter was not correct. The disc isn't thin provisioned. ».

  It seems I have dealing with the same thing, as this is the document here (http://pubs.vmware.com/Release_Notes/en/vsphere/55/vsphere-vcenter-server-55u3-release-notes.html).  However, I have no idea what to do now... and I'm worried because it seems that my backup plan may not work if I can't restore an OVF/VMDK from a disk of visas.

  The solution to the first problem here: https://communities.vmware.com/message/2172950#2172950 that solved my problem 2nd too.

• Save space on the disc thin provision

  Hello.

  I have a virtual machine that has thin provisioned N. 3 disks:

  • 1-> provisioned size 20 GB
  • 2nd-> size 3 GB provisioned
  • 3-> provisioned size 20 GB

  If I look on the use of the storage on the server ESXi 5 (VI client) I see this:

  

  But if I connect on the virtual computer, I see this:

  

  With a simple calculation, I don't see that really used on the virtual machine storage size is only (20 - 8,64) + (3-0, 746) + (20 - 5,46) = 28,15 GB.

  Very different 40,48 GB storage used .

  The question is:

  is it possible to reclaim unused storage space?

  In other words: is it possible to align the "storage used" showed on the resources of the ESXi server to that actually used from the VM?

  Thanks in advance.

  You can use vmkfstools with the option ofpunchzero to recover the disk space of the thin configured virtual disks. To do this, you must first set to zero unused space within the OS itself invited, then turn off the virtual machine and run the vmkfstools command.

  André

• Thin provisioned VMDK consume all available space

  Hello

  • Server: Win 2008 R2 with SP1 (file server)
  • Storage: VMDK on VMFS5 datastore (thin provisioned), which underlies the SAN fiber storage
  • vSphere: vCenter Server Standard 5, ESXi 5.0 Patch 2 (build 515841)

  ... So, I have a VM file server running Windows 2008 R2 with SP1.  The system (c) partition has been successfully thin (VMDK using 13 GB to 40 GB).

  However, for all subsequent hard disks virtual created on the server, the NTFS partition is very slow to get into shape - and the actual size of the VMDK creeps slowly up to the size of the VHD (you can see the gradual rise of the VMDK size in the browser to store data during the formatting process).

  The VMFS data store that we create new virtual hard disks is version 5.  All virtual machines with virtual hard disks on VMFS version 3 seem to be supplied thin OK.

  Nobody knows what could happen here?

  Hello

  Do you use quick format?

  Otherwise, your score will take up any space. Check here:VMware KB: using thin provisioned disks with virtual machines

  "If a guest operating system needs to use a virtual disk, the must invited first operating system partition and format the drive in a file system it can recognize." Depending on the format selected in the guest operating system, the format can cause the disc thin provisioned in full size.

  For example, if you have a drive end with a Microsoft Windows operating system implemented and format the drive, unless you explicitly choose to fast formatting, Microsoft Windows format tool written information to all sectors on the disk, which in turn inflates the disc thin provisioned in full size. "

  Kind regards

  Julien.

• Exchange 2010 - is thin provisioned supported or only zeroeager?

  It seems to be conflicting information anywhere on the internet.  When configuring storage for the newspapers and the Exchange databases, disks must always be zeroeager or are provisioed thin discs supported these days in VMWare?  MS say this extension dynamic disks are not supported, but which is under Hyper-V

  See attached PDF.

  Q: are there requirements around storage for my virtualized Exchange solution design?
  A: VMFS on iSCSI and Fibre Channel storage is supported, as are the physical maps and virtual-mode raw device. The use of thin provisioned disks or network - storage devices is not supported by Microsoft for virtualized Exchange deployments.

  Confirmation by Microsoft;

  http://support.Microsoft.com/?kbid=897615

• Ability to store data when thin provisioned

  Hi all

  VSphere 4.1 running in a cluster of two hosts, this cluster uses a Dell Equalogic PS4000 SAN for storage.

  Confusion I have is with thin configured data stores and my Exchange Server (2010) - running on MS Server 2008 R2 Ent.

  Exchange server uses three data warehouses on the San, one for the OS, one for data and one for newspapers.

  The data of databases store is the one that gives me problems, it has been defined as thin provisioned 750 GB, when I look on the SAN & via VCenter he said 90% full until she went to 100% after the SAN sent several alerts - the virtual machine then crashed with out-of-space error.

  To cut a long story short, I have twol questions

  (1) in this case why the SAN & VMWare say the data store is full, BUT when I connect to the virtual Windows computer it was about 50% in use?

  (2) when the data store that contains the databases filled I don't understand Exchange has fallen more but why why the crash of the VM?  The OS data store still had plenty of free space.

  I got this error by increasing the space of 100 GB to 850 GB data store while the virtual machine is very well (even if Windows was complaining to a stop suddenly unexpected).

  Thanks in advance for your help.

  K

  And without perspective, as it was put in service end and touch 750 GB, 300 GB, then was removed, it would not 750Go because it may not decrease in size - is that correct?

  (This would answer why the SAN says he uses 750Go of 750 GB, but Windows thinks there ish 300 GB free).

  Yes, that's correct... the recovery of unused/dead/removed space don't is not automatically made.

• Thin provision NFS works only for instant VMDK

  Hi guys,.

  We have a host of ESXi 4.1 using NFS as a storage device. NFS is provided by the platform, Bluearc Mercury.

  Whenever I create a new virtual machine, for Thin Provisioning option is checked but also gray outside. There is a line saying:

  "Actual allocation policy will be determined by the NFS server.

  I'll check the newly created VMDK and it's a thick disk, it consumes as much disk space I put for the disc size.

  So, it makes me believe that the Bluearc Mercury does not support provisioning.

  The strange thing happens after that I did a snapshot, another VMDK as server - 0001.vmdk was created, and it's the thin disk.

  Someone at - it the similar problem?

  Kind regards

  Derrick

  Showmanlkz,

  On the BlueArc server, file systems are provisioned files, thin as VMDK files are created as "fragmented files".  Just a slight difference in terminology, but the effect is what you're looking for in what concerns the commissioning only the disk blocks that are actually used (thin provisioning).

  If you have created a 10 GB VMDK file, but have only 5 GB of data on the inside, only pointers to these 5 GB of data blocks will be created.  Us will not pointers to blocks of empty data.  Standard orders will make reference to the metadata VMDK structure and will report on the 10 GB file.  In order for a command 'df' to show the use of real drive (5 GB), rather than the allocated size (10 GB), you must enable the "true fragmented file" mode.  Contact the BlueArc TAC for the real run command.

  Snapshots are a completely different beast.  By default, they are only pointers to a point static instant copy.  However, if you use our new feature JetClone to create clones in entry level file, these files will be also thin provisioned, only as the clone diverges in content data from the original VM has been created between more and more.

  Julie HG

  Product Manager, BlueArc

• Thin provisioning question

  Tonight

  We recently P2V. by 4 Dell servers that all went well.

  When it came to thin provisioning of storage, we encountered a slight impediment. No matter what we do, we cannot recover any space with a provisioning.

  For example, we have only one server, 62 GB C. We have done the following:

  Defragmented

  CHKDSK would be

  SDelete - c (and even z, followed by another - c)

  Then migrate to another store of data configured as thin. We seem to get 0 space back even if Windows reports being 31.6 GB of free disk space.

  Don't get me wrong in the past doing the 4 steps above still retrieved the free space, but just this time on all 4 servers, we cannot recover anything.

  I have even converted to thick, and then to thin in case something to be funny.

  With the help of ESX 4 fully patched on all hosts. The only thing I can think is that when we thin put in service in the past, we were running ESX 4 with no patches and we have recently updated to the current version, but I can't really believe that a fix/update would have broken something as Basic.

  Any idea?

  There is a very simple solution for this problem, I found a couple of weeks ago.

  Do the usual "sdelete - c C:" stuff, and then migrate as Thin put in service in a store of data with a * different blocksize *.

  Then it just works how it has always been.

  It's only when you migrate it to a store of data with the same size block, it's broken.

  --

  Jules

  [email protected]

  www.Jules.FM

• Thin Provisioning vs thick Provisioning

  Some users say they are not too Committee a disc thin provision.

  But I think that if they are not over - commit then how are they case using the benefits of provisioning. Instead why don't they use thick available?

  You guys agree with my opinion?

  I agree, if there is enough disk space, thin provisioning is basically no benefit to this transfer.

  André