VMware Cloud Community
mldmld
Enthusiast
Enthusiast

DR - Resignature or not resignature of a datastore : that is the question

Hi all,

I plan to install a vSphere 4.1 cluster on 2 close sites: Site A and Site B. N ESX on each site. All ESX are in the same VLAN with same subnets.

Some VM will be stored on datastores on Site A. Some VM on datastores on Site B.

The datastores of theses VM will be mirrored by the Storage Arrays from Site A to Site B, and Site B to Site A.

I don't know SRM and I would like to avoid it this year.

The clone datastores are read only. So they can't be mounted by ESXs as far as I understood.

In case of a DR where storage on one site is gone :

=> Storage administrators make clones writable.

=> When vpshere admin mount the clones, vsphere ask about datastore signature because the  VMFS volume signature do not match, because of the Lun Number/disk serial  number are not the same as the originals datastores.

So what should  I do

- Mount the datastores with their existing signatures ?

- assign a new signature ?

What are the pros and cons if I have 1 cluster splitted on both sites or 2 clusters, one on each site ?

Is there a way to avoid to register all lost VM from the cloned datastores ?

In case of a single cluster splitted on site A and site B.

=> I will setup 2 hosts group where VM stored on site A should run on site A, and VM on site B should run on site B

=> All ESXs see datastores of Site A and Site B

Pro :

-CPU load can be shared on all ESX of the cluster if needed.

-if I loose all ESXs on a site, VM will be restarted automatically by HA on the remaining site without activate de DR plan

Cons

?

If I create two clusters on each site A and B :

- ESX of cluster A should see datastores of site A and datastore clones of site B

- ESX of cluster B should see datasotres of site B and datastore clones of site A

=> As clones are read only, it is still impossible to mount clones

=> In case of DR, the storage admin make clones writable.

Pro

?

cons :

- if I loose all the ESX of a site, but the storage is ok, I must activate the DR plan.

Thank you

Best regards

0 Kudos
1 Reply
AndreTheGiant
Immortal
Immortal

You can keep the existing signature.

If you use snapshots of the clone than could be better use a resignature.

If you have clone and/or snap, you cannot have the same cluster across the two site.

One cluster must see the SAME datastore in order to have vMotion, HA, DRS, ... working.

Also to have same cluster you need a stretched LAN across the two sites, and this could not be possible in some cases.

So I suggest two clusters.

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
0 Kudos