asked on

VMware DataStore clarifications

Hi Experts ,

I need some clarification about VMware Data Store :

1- our application needs to have 3 Drives (C:OS ,D: Datasbase ,E:Log) I can achieve this by creating the first disk when preparing my Virtual machine and then go to add disks then select different datastore is this a proper way and best practice ? cause when I am looking into it the virtual machine files separated into three different data stores and different raid Level as per the applications need.. will that affect the application performance ?is there other way to do this ?
2- will be a difference if I create the data store from the cluster level or the host it self ? as I can see all hosts can see the same data stores even if I created it from the host level
3- what is the difference between unmounts and de attach for datastore?
4- what the proper way to delete a datastore completely ? I have made a test one and when deleting it says force mounted VMFS datastore will be removed from inventory but data on them will not be deleted so do I still need to do anything form VMware side ?or just delete the Storage Lun?
5-when creating data store should I give it the maximum capacity given by the storage ? yet it still can access the full but when u read capacity it will show what you select full or less
6- what is the difference between a datastore and a cluster data store?
7-if I created a datastore and select the default block size (1 MB ) can this Be changed later on or no ? as I didn't find the option and what is the recommendation block size for Oracle DB

thank you

Andrew Hancock (VMware vExpert PRO / EE Fellow/British Beekeeper)

1- our application needs to have 3 Drives (C:OS ,D: Datasbase ,E:Log) I can achieve this by creating the first disk when preparing my Virtual machine and then go to add disks then select different datastore is this a proper way and best practice ? cause when I am looking into it the virtual machine files separated into three different data stores and different raid Level as per the applications need.. will that affect the application performance ?is there other way to do this ?

Creating a disk per Volume (Drive Letter) is Best Practice, to avoid legacy partitions, each volume is easier to extend, and you've also hit the nail on the head, it allows you to locate the virtual machine disk, on different VMware datastores, which could be different tiered storage, e.g. spinning rust, SSD or NVMe flash.

If you place different virtual disks on different datastores, it can affect the Application performance, e.g. RAID selection, Disk Speed, e.g. 7k, 10k, 15k, SSD can all effect the performance based on IOPS the datastore can provide.

2- will be a difference if I create the data store from the cluster level or the host it self ? as I can see all hosts can see the same data stores even if I created it from the host level

No difference, all datastores are attached to ESXi hosts.

3- what is the difference between unmounts and de attach for datastore?

Unmount - removes the datastore, de-attach - detaches the LUN.

So you unmount the datastore first, and then detach the LUN.

4- what the proper way to delete a datastore completely ? I have made a test one and when deleting it says force mounted VMFS datastore will be removed from inventory but data on them will not be deleted so do I still need to do anything form VMware side ?or just delete the Storage Lun?

see 3 above. (or you could select Delete, then dettach)

and here

https://kb.vmware.com/s/article/2004605

5-when creating data store should I give it the maximum capacity given by the storage ? yet it still can access the full but when u read capacity it will show what you select full or less

Create the LUN size you require, and then use the full size to create your datastore, you can always expand LUN and Datastore later if required, do not use all your SAN storage and allocate it in advance.

6- what is the difference between a datastore and a cluster data store?

A Clustered Datastore, is a group of datastores in a Cluster, which can be used for SDRS, e.g. virtual machine disks storage will be Storage vMotion around the Cluster Datastores, just like DRS moves VMs (compute) around hosts in the cluster.

7-if I created a datastore and select the default block size (1 MB ) can this Be changed later on or no ? as I didn't find the option and what is the recommendation block size for Oracle DB

VMFS Block size is 1MB and cannot be changed.

Ref: Oracle DB

see pages 36, 37, 38

https://www.oracle.com/technetwork/server-storage/sun-unified-storage/documentation/partitionalign-111512-1875560.pdf

mu3tasem

ASKER

Thank you Andrew ,

Regarding Point# 1again

is what I am doing is best practice ? to create disk for each drive , is there another way to achieve this ? to make multiple drives

in my case OS will be in raid 1 , DB on RAID 5 , Log on RAID 1/0 from 3 different data stores ,shall I continue to build based on this setup or it will be better if make a single data store ?
Will it make the backup slower if I backup the virtual machine in my scenario ?

ASKER CERTIFIED SOLUTION

Andrew Hancock (VMware vExpert PRO / EE Fellow/British Beekeeper)

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial