C3000,Lefthand and V-Sphere

[DThomson_1]

Hi

Im totally new to the Lefthand VSA technology and i have a few questions, if any once could help.

Ive read a few things here on the forum and im confused.

This is what i was hoping to implement:

C3000 Blade system
2 x BL460c Server blades
1 x SB40 Storage blades
LeftHand P4000
Vmware VSphere - HA/DRS/VMotion

I want to build 2 server blades into a VM Cluster using one of the SB40 storage blades as shared storage for the Virtual servers.

My concern: Will HA/DRS and VMotion work if one of the servers fail ?

[DThomson_1]

Sorry, thats the Lefthand software, not the device.

[Uwe Zessin · Accepted solution]

You cannot have automated failovers with two VSAs/storage modules only. You need to configure a 'virtual manager' for manual failovers or install a 3rd VM on a 3rd computer that runs a 'Failover Manager' (FOM).

There are two versions of the FOM than can run on ESX or VMware Server.

[DThomson_1]

Thank you Uwe.

[Steven Clementi]

Isn't the SB40 only directly accessible to one blade at a time though? (The adjacent blade)

Just trying to understand...

You are going to set up 1 server, attach it to the SB40, install ESX on that server, implement a LHN VSA on the one node... so that you can present the storage... via iSCSI... to both of the ESX servers?


Do you see yourself getting additional blades and SB40's in the future? so that you can expand the LHN storage with additional "nodes"?




Steven

[Uwe Zessin]

Good point. To cover the failure of any server blade it is necessary to set both up with:
- local storage
- a VSA
because VSAs do not work with shared storage. Storage redundancy is then created by doing 2-way replication (think of it as a mirror with 2 members) between VSAs.

Maybe it is not obvious, but: if you have only two ESX servers and one of them fails, you cannot do VMotion or DRS (which uses VMotion). HA, however should be able to restart (=reboot) the failed VMs on the surviving server as long as enough resources are left.

[DThomson_1]

Hi Steven

As i said im new to this Lefthand tech.

The storage is only accessible from the adjacent blade but with LeftHand you can share the storage across more blades in the enclosure. This is what i understand about the tech. Good for shared storage in an ESX environment but you cant have HA and DRS without another server running FOM.

Im now looking into an MSA to use as external storage now as the project has expanded quite a bit since yesterday.

[DThomson_1]

Morning Uwe

I cant seem to find reference to Vmotion and DRS not working between 2 ESX servers.

[Uwe Zessin]

You asked:
> Will HA/DRS and VMotion work if one of the servers fail ?

My response is that VMotion/DRS will not work after one out of two servers has failed. You no longer have a destination server that you can migrate a VM to.

[DThomson_1]

Sorry, yes if one of them fail.

If there is enough resources on the second server the VM's should re-start their ?

Im searching the web and i cannot seem to find a requirement of 3 servers minimum for a working ESX cluster that uses HA DRS and Vmotion.

I automatically assumed that 2 would suffice if their was enough resources.