1753776 Members
7051 Online
108799 Solutions
New Discussion юеВ

Left Hand P4300

 
teledata
Respected Contributor

Re: Left Hand P4300

You certainly can run on the production ESX boxes. Just make sure you use local storage for the FoM.

The reason I sometimes elect to use ESXi is as follows:

Let's say you have a power failure, your ESX boxes reboot, they take a long time to timeout connecting to your Lefthand iSCSI targets. Your FoM can't come online until your ESX box completely finishes the boot process.

The ESXi box using local storage only can boot quickly, and launch the FoM thus getting your quorum re-established faster.

Plus I usually have some leftover hardware from a virtualization consolidation project, and have some management tools VMs running on an ESXi box anyway, so It's certainly not a requirement, but I like the recovery design of using an off-cluster ESXi utility host.
http://www.tdonline.com
rdsworks
Occasional Visitor

Re: Left Hand P4300

I'm new to HP SANs and am trying to configure something every similar to what you mention here.

 

I have two HP P4300 in two sites, in order to have each site work independently if the fiber is cut would I need to create two clusters?

 

Which method would be best to use for having the data available at both sites?

 

Thanks,

 

Dan

oikjn
Honored Contributor

Re: Left Hand P4300

You can never have two copies of a LUN available on two different sites while the two sites aren't in communication with eachother.  Ask youself how it should handle a situation where one site writes anything to the LUN and then the other site writes something else to the same spot!

 

read up on the DR manual HP provides as that should help with your planning.  With two independent clusters, you can always do failover/failback and/or read-only access while the primary is available.  For a single multi-site cluster, there are other complications, but if the link goes down ONE site will work and the other won't, you can design which will stop, but one simply has to stop to prevent data corruption of the cluster and corrutption is more of a problem than uptime is.