Storage VMFS Partition Performance

Similar Questions

• It is possible to expand a vmfs partition?

  Well, here's my problem, I have 2 esx servers, connected to DS3400, people who originally this set up only bought 4 disks and created a raid 5 1.6 to VMFS partition. The guy he filled and I was running out of space.

  So I bought 2 more disks. I added the readers to the controller and added to the raid, creating a bigger raid.

  I returned in vsphere and saw that there was free space and I tried to extend VMFS partition but I couldn't? I thought that I was not supposed to create a new measure.

  In my frustration, I said F and I just created a second logical drive, formatted as VMFS and added and moved some of the VMm on the new 800 GB partition.

  Now, I bought 4 records more than daughter upward the ds3400. I'm trying to add them to the table.

  I need to make the a VM on the origial 1.6 TB of greater. So I don't want to add a third partition, instead of this I would really be able to extend the partition of 1.6 to 3.2 to with new readers, I add.

  Does anyone know how or if I can do this? Its going to take some time to add the new disks and I need the extra space pronto.

  When I'm in the IBM Storage Manager I see

  Unconfigured space 1.2 to (always adding these readers)

  RAID 5 2.8 TB

  -Original DS (1.6 TB)

  -1 (837 GB) second one I did

  -Release the 418GB capacity (one of the 3 new discs add)

  Once I get the other 3 dirves, I'll have another 1.6 to release capacity on the raid. This is where I'm stuck, how to use that to extend the original 1.6 to vmfs.

  Thank you people a bunch.

  It is possible to expand a VMFs partition - but the problem is that you can not access a logical unit number is greater than 2 TB - 512 B and the largest virtual disk is the same 2 TB - 512 according to http://www.vmware.com/pdf/vsphere4/r40/vsp_40_config_max.pdf - using extensions you can create a volume of 64 KB-16 TB per individual LUN stacking

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

  Session data >

• Extend VMFS partition alignment vs. partition

  I have a (openfiler) iSCSI storage and a host ESXi (running on its local disk) for testing.

  I tried to create VM customers and together start iSCSI. The problem I have is the alignment of partition and extend VMFS partition.

  What happens:

  If I want to align parititon, I need to go up and format the LUN with the VMFS. Therefore, the VMFS partition size is fixed. I'm able to increase the size of the LUN, but I can't do anything with regard to the size of the partition resizing VMFS.

  Another way is to report the storage for the VM client on raw storage. But this way, I can't align the partition because I can't use fdisk on raw storage on CLI.

  My question is: if I want to run SQL server, for example, and want to align the disk for the storage of data and able to rescale the size, how can I do?

  Thank you.

  I don't know exactly what you're asking, but "growing" a VMFS volume does not resize the partition, it is done by adding one "scope" (another partition) and then ESX (i) of the volume manager puts the two partitions together to make a large volume.

• The VMFS partitions and ESX in MUV upgrades

  Back in the day when you installed ESX it deems any VMFS partition is priese formatting for you even if you had all your machines on them

  I intend to upgrade ESX 4.0 to 4.1 using AUVS that resembles a simple process.  Anyone know if:

  1. it clears the local VMFS partition?

  2. He wipes SAN VMFS partitions that it finds?

  Thank you

  Stu

  N ° you don t be vmfs http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1022140 loss

• expand a vmfs partition running

  I have a HP EVA 4400 iSCSI, which has 2 partitions vmfs (vdisks).  It is formatted with a block size of 1 MB, the other is the block size of 8MB.

  My ultimate goal is to have all the vm that are on the score of 1 MB on the partition of 8 MB, blow the partition of 1 MB and while space on a partition of 8MB.

  I never increased an existing VMFS partition, so I don't know what to expect (I've never had need of).  Is it as simple as deleting (using 4.1) 1 MB one, adding that the space in the vdisk the 8 MB one and then saying ESXi to increase?  It will use anything he can find, or what I have to tell him how to increase by?

  Are the virtual machine that will be on the secure partition of 8 MB?  I don't have a quick way of enclosing all down and save them.

  Any help would be appreciated.

  The way I read your message, it's that you want to migrate to the size of block 8MB data store, delete the data size of block 1 store MB and have its place to extend the data store of 8 MB for a larger size.    If that's what you want to do then you should give this blog post read, http://www.yellow-bricks.com/2009/03/26/resizing-your-vmfs-the-right-way-exploring-the-next-version-of-esxvcenter/.

  If you have found this device or any other useful post please consider the use of buttons useful/correct to award points

• Convert VMFS partitions/data?

  Hi again!

  IV been playing with Esxi, Hyper-V and Xenserver for some time now. Finally, I decided to go with esxi for my virtualization project but there are 2 questions, I need to get an answer first.

  1. None of my raid cards that I have at the moment are detected by esxi. I read that theres a lack in the nucleus of esxi so that you can't do software raid; is that correct?

  2. I 3xTB drivers where partitions are encrypted with truecrypt, what is the easiest way to convert these VMFS partitions so I can use them in esxi?

  Thanks for your help!

  Hello

  As far as I know, you have two main options:

  (1) create a virtual machine with a normal vdisk (hard) to place the data of truecrypt. Then make a transfer to network on the virtual machine to a system that has access to the unencrypted data. Finally the date of stay to the virtual machine without encryption vdisk

  (2) create a virtual machine and install the truecrypt sotfware. After that add truecrypt disks (perhaps is better to merge all data from a larger drive?) to the esxi machine. The esxi should see the disk (s) connected (s) as LUN SCSI/SATA. Then, create the physical (s) or the discs as raw device mapping (RDM). In my humble VIEW in your scenario is preferable to create RDM in physical mode. Finally using the truecrypt software you should be able to access the encrypted data.

  Here, you can check a few RDM tutorials:

  http://www.avatir.com/HOWTO-RDM-on-local-SATA-disks-in-ESXi/

  http://VM-help.com/esx40i/SATA_RDMs.php

  http://ServerFault.com/questions/105652/assign-and-remove-multiple-LUNs-from-VMs-on-ESXi-4

  It may be useful

  A saludo/best regards,.

  Pablo

  Please consider providing any useful answer. Thank you!! - Por favor considered premiar las useful responses. MUCHAS gracias!

• Extend an existing VMFS partition

  Is it possible to increase the size of a VMFS partition without using extensions in ESX 3 U4?  Everything I see in the manual server configuration is to use extensions.

  Adam

  No, 3.5 cannot extend VMFS partition. The only way - create new old size desired, migrate all virtual machines and delete data store.

  ---

  VMware vExpert 2009

  http://blog.vadmin.ru

• ESXi 3.5 - clean installation, it will wipe existing vmfs partition?

  So let's say I have a U3 3.5i ESXi server is installed on a single table of RAID 5.  I want to do a clean reinstall using U4.  When I installed U3 it parititoned discs automatically, so when I go to install U4 will be it preserve the current vmfs partition or is it delete and then recreate a new?

  The reason I ask this is just in the preparation of a time where a firmware update could possibly fail (do not assume that it will be, just want to have a plan in case something goes wrong).  Just want to be sure I can do a reinstall with/without losing existing virtual machines.  I'm not worried about having to reconfigure everything - not a big deal.

  Thank you!

  You have the ability to clean any vmfs partition, or leave them alone.  It is up to you to destroy the existing, which you do not want if there's vm is here, or leave them alone.

  -KjB

  VMware vExpert

• All gone VMFS Partition tables! on LUN

  Today, we had a 20 TB of data problems and all partition tables VMFS got blownen now. All ESX servers were not able to see the data stores. Everything is back to normal now, but I have a question, not even a VM has stopped working? How is that possible? VMware said eventually they would have all the blue screen, but the LUN was down to 2-4 hours and not even a VM is down, we have more than 300 + VM can someone explain how it worked?   I guess they were running in the memory cache and depending on the amount of data is being written depends on when they started bluescreen? How does it work?  I am pleased, that was the case...

  Hello

  Discover the recovery VMFS partitions for assistance. SMV will stay running because they know where they write on the disc through a caching mechanism. The disk data were not deleted, only the partition table.

  Best regards
  Edward L. Haletky
  VMware communities user moderator, VMware vExpert 2009
  ====
  Author of the book ' VMWare ESX Server in the enterprise: planning and securing virtualization servers, Copyright 2008 Pearson Education.
  Blue gears and SearchVMware Pro items - top of page links of security virtualization - Security Virtualization Round Table Podcast

• VMFS partition alignment and recommendation of the storage provider

  In the literature of my storage provider it says they use a band of 256 KB size, and it's a good approach to set the offset to 256 KB (or 512 sectors) when creating VMFS volumes.

  "When you use the VI Client, the new partition is automatically set to an offset of 128 sectors = 64 k." But,
  in fact, this configuration is not ideal when you are using the DS8000 storage disk. Because the
  DS8000 uses more large block sizes, the offset must be set to at least the size of the stripe. For
  RAID 5 and RAID 10 in the attachments of open systems, the allocation size is 256 KB, and it's a good
  approach to set the offset to 256 KB (or 512 sectors). You can configure individual compensation
  that from the ESX/ESXi server command line. »

  I'm guessing that they always use the old VMFS3 information above?

  Now with vSphere 5.x and VMFS5 the offset is to the sector 2048 (or 1024KB)

  Question is if I go with vSphere 5.x VMFS5 and create the volume VMFS via vCenter VI Client will be I'm always perfectly aligned with my storage or should I still manually create VMFS volumes with starting with sector 512 (256 KB)?

  Thank you for your answers!

  1 024 KB alignment is a multiple of many sizes of block by default if it guarantees a correct alignment for all types of storage where the result of 1024 / 'stripe size' is an integer value.

  André

• Storage partitioning - query on multiple partitions performance?

  Hello
  
  We use Oracle 11.2.0.3 and have a central fact table large with several alternative identifiers that have the index bitmap on them and fks looking at the dimension tables + several measures
  
  (PRODUCT_ID,
  CUSTOMER_ID,
  DAY_ID,
  TRANS_TYPE_ID,
  REGION_ID,
  QTY.
  VALUE)
  
  We have 2 separate sets of queries users seek to run for the most part, those accessing transactions products regradless of the time transactions that happened (that is, not financial queries - approximately 70%,)
  queries to determine what happened in a particular week - 20% of the queries.
  
  Table will be about 4 billion lines eventually.
  
  Whereas the addition of additional column to that partition the DATE range and this for us to abandon the old partitions each year - but these data would not be joined to any other table.
  
  Whereas then infra partitioning by hash of the product_id that is the key to override to the product dimension.
  
  Reflections on the performance?
  
  Queries by their nature would have hit several secondary partitions.
  
  Reflections on the performance of the queries of queries that access several sub-sub-sheet music/scores compared to applications running on a single table.
  
  Any other thoughts on partitioning strategy in our situation much apprecaited.
  
  Thank you

  >
  Hash partitioning gives not the size of partition with these predicates to join because they use the equality operator?
  >
  Size of the partition occurs at the stage of the analysis. Oracle analyzes the query used and for partitioned tables, trying to determine which partitions are necessary. If filter predicates include information which indicates that one or more specific partitions the Oracle can "prune" other partitions because they are not necessary.

  So if 'product_id' is partitioned hash then all rows with the same value will be in the same partition. Then a query like

  SELECT * FROM myTable
  WHERE PRODUCT_ID = 7
  

  Oracle will know when the query is analyzed which sharing one and one only, is necessary because the value 7 always hash to the same partition and Oracle can chop the literal "7" at the time of the analysis to determine which partition is necessary; so cut the other partitions.

  If product_id is used in a join predicate to another table but is not limited to specific values like the example above then Oracle cannot determine at the time exactly what partitions might be necessary because it will depend on the discovery at the time of actual data values.

  The information needed to trim must be available at the stage of the analysis. In other words, the query and the metadata for the table (DDL, statistics, etc.) will be used to make decisions of pruning - the actual table data are not a factor because the data are not available at the time of the analysis.

• VMFS Partition alignment

  We have a small and somewhat new virtual environment.  Most of the clients are windows with a bit of Linux.  Now hear there is a partition with VMFS alignment problem regardless of storage provider.  Can anyone offer any indication as to whether this should be a matter of concern (performance in mind) - should the guests we have already ctreated be realigned?  We have other physical servers connected to an EMC SAN, and even if we changed the starting at 128 block was not any différenec in performance.  If realignmemnt is recommended then what is the best way to do it?

  See also:

  Disk partition alignment

  André

• VMFS Partition 5 alignment

  Hello

  Am I right in thinking that now with 5 VMFS using the unified 1 MB block size and as long as you create the data through the viclient store it is not necessary to align the partitions as you used to have to ESX3, 4?

  Thank you

  Gregg

  Cormac Hogan wrote an article on this subject:

  http://blogs.VMware.com/vSphere/2011/08/vSphere-50-storage-features-part-7-GPT.html

• Retrieve VMFS partition table... help.

  Hi all

  This morning, when I boot ESXi 4.0.0 I found lost a VMFS volume. By default, my drive has two partitions. I tried to recover the second partition with the instructions on the forum, but it was still not working.

  fdisk- lu/dev/disks/t10. ATA___ST3500418AS___9VM6Z8Y1

  Disk/dev/disks/t10. ATA___ST3500418AS___9VM6Z8Y1: 500,1 GB, 500107862016 bytes
  255 heads, 63 sectors/track, 60801 cylinders, total of 976773168 areas
  Units is 1 * 512 sectors is 512 bytes

  Device boot start end blocks Id system
  / dev/Disks/T10. 63 8386622 4193280 6 FAT16 ATA___ST3500418AS___9VM6Z8Y1p1
  Partition 1 does not stop the limit of cylinder
  / dev/Disks/T10. ATA___ST3500418AS___9VM6Z8Y1p2 8386623 976773167 484193272 + fb VMFS

  
  fdisk u /disks/dev/t10. ATA___ST3500418AS___9VM6Z8Y1

  The number of cylinders for this disk is set to 60801.
  There is nothing wrong with that, but it is higher than 1024,
  and could in certain setups cause problems with:
  (1) software that runs at startup (for example, the old versions of LILO)
  (2) start and partitioning software to other operating systems
  (e.g. BACK FDISK, OS/2 FDISK)

  Command (m for help): d
  Partition number (1-4): 2

  Command (m for help): w
  The partition table has been changed!

  The call ioctl() to re-read the partition table

  u Fdisk/dev/disks/t10. ATA___ST3500418AS___9VM6Z8Y1

  The number of cylinders for this disk is set to 1409485.
  There is nothing wrong with that, but it is higher than 1024,
  and could in certain setups cause problems with:
  (1) software that runs at startup (for example, the old versions of LILO)
  (2) start and partitioning software to other operating systems
  (e.g. BACK FDISK, OS/2 FDISK)

  Command (m for hELP): n
  Control action
  extended e
  p primary partition (1-4)
  p
  Partition number (1-4): 2
  First of the sector (by default 8386623-976773167, 8386623): using default 8386623
  Last sector or size + or + sizeM or + sizeK (8386623-976773167, by default 976773167): using default value 976773167

  Command (m for help): t
  Partition number (1-4): 2
  HEX code (type L to list codes): fb
  Changed the type of partition 2 to fb (VMFS)

  Command (m for help): x

  Expert command (m for help): b
  Partition number (1-4): 2
  New beginning of data (by default, 8386623-976773167, 8386623): 8386688

  Expert command (m for help): w
  The partition table has been changed!

  Calling ioctl() to re-read the partition table.

  I have VL connection and go to the Configuration > storage > refresh the volume does not return.

  Please help me!.

  This seems really odd. Do you have all the partitions on these 4 discs when you first installed ESXi?

  What version of ESXi you originally installed (3.5 or 4.0) in the case where that this system has been upgraded?

  Please after the release of fdisk-lu for all 4 disks in the system.

  André

• vs VMFS mapped performance raw lun

  Here are two tests that I ran inside a virtual machine.  One is on the OS disc that is in a regular VMFS data store, the other is to a raw lun mapped.  Both are fixed on the same back-end (Equallogic iSCSI) storage connected via MPIO 1 Gbit/s.  Should I see this kind of difference and 37 MB/s OK?

  # dd if = / dev/zero bs = 9000 count = 100000 = test & & rm-f test

  100000 + 0 records in

  100000 + 0 records out

  900000000 bytes (900 MB) copied, 8,83205 seconds, 102 Mbps

  # dd if = / dev/zero bs = 9000 count = 100000 = test & & rm-f test

  100000 + 0 records in

  100000 + 0 records out

  900000000 bytes (900 MB) copied, 23,9424 seconds, 37.6 Mbps

  > I can't imagine which translates into the difference between RDM and VMFS

  Difference between RDM and VMFS is negligible. The significant performance of vmdk less - reset default records (due to security problems).

  ---

  MCITP: SA + WILL, VMware vExpert, VCP 3/4

  http://blog.vadmin.ru