LUN to Datastore

[aymanshawarby]

VMware since version 4.0 recommended to use LUN size of 2TB. Is there is any new updates about the LUN recommended size and best practice ?

Also is it recommended to use 1 LUN for 1 Datastore, or add more than 1 LUN in the same Datastore. Since when create a Datastore with more than 1 LUN and create machines. VMware start to write on the first LUN until it's full and then go to the second LUN. that's mean it is concatenated  not stripping technic. This can cause a performance issue if there is applications with High IOPS.

Need your support and advice. If you can share with me vmware documents about the best practice this will be helpful.

[PCTechStream]

Best practice is dependent upon your environment. You won't necessarily find best practices on datastores necessarily. Datastore configuration depends most likely upon:

1. How many VMs you have

2. Type of VMs you have (e.g. IIS, MX, SQL, etc.)

3. Performance requirements, etc.

Your LUN disk RAID array for the datastore will determine performance, as well as how many VMs and the type of IIS, VMs (MX, SQL, etc.) on those LUN disks. So, as you can see, it really takes a LOT of brainstorming on how best to set up your virtual environment.

EXAMPLE:

My "best practice" 500GB LUNs for ESX/ESXi on FC SAN. Typically I may have 8 or 9 hosts per LUN.

Check this VMFS Best Practices

LINK: http://www.vmware.com/pdf/vmfs-best-practices-wp.pdf

Also, VMware VMFS Datastore Sizing Calculator

LINK: http://thinkingloudoncloud.com/calculator/vmware-vmfs-datastore-sizing-calculator/

Raul.

VMware VDI Administrator.

http://ITCloudStream.com/

Stay Connected: Like > Share > Comment > Reply > Helpful > Answered

[vfk]

1 LUN 1 Datastore is best practise, concatenated LUNS are called extents in VMware documentation.  Here is a good link - VMFS Extents - Are they bad, or simply misunderstood? - VMware vSphere Blog

Datastore LUN can now be as large as 64TB, you should size your LUNS according to your needs, along with RTO / RPO capability in the event you lose a LUN for whatever reason, i.e corruption.

You should consider Storage DRS for pooling datastore, you can set usage and performance threshold etc.