|
|
|
Forum Newbie
Group: Forum Members
Last Login: Thursday, October 25, 2007 3:18 PM
Posts: 2,
Visits: 5
|
|
I am looking for some clarity in how this all fits together. Maybe if I just did it things would make more sense but some confirmation from the pros would be nice.
ESX server boots up. Next the VSA has to boot to make the iSCSI target available so the rest of the VMs can boot up. Lets say I have two 2T DL380 G5 servers to play with. What are my drive configurations? Hardware RAID or not? I would plan on RAID1 on the VSAs I would think. So what is the best practices plan for drive layout in ESX for these server to utilize VSA? How about once VSA was installed?
|
|
|
|
|
Supreme Being
Group: Moderators
Last Login: Yesterday @ 3:37 PM
Posts: 108,
Visits: 375
|
|
ESX server boots up. Next the VSA has to boot to make the iSCSI target available so the rest of the VMs can boot up Only the case if you have only a one node VSA cluster and the VMs on that exist in the same ESX server. If you are doing that I'd assume it is so you can use asynchronous replication or some other SAN/iQ feature on the VMs. All you'd have to do is specify the VM boot up order to have the VSA first (a best practice in the manual) and list the other VMs to boot up later or do them manually. Either way the ESX server is going to take a while to boot up because it will look for iSCSI targets at boot up. More typical would be to have a many node cluster in which case the iSCSI sessions and VMs hosted in them don't have to wait. They can connect to the other live VSA nodes in the cluster. Once the VSA on that ESX server is booted up it will join its cluster and re-sync up, that does not have to happen before the other VMs can boot though. What are my drive configurations? Hardware RAID or not? Hardware RAID (another best practice listed in the manual) is advised. Otherwise you'd have to rebuild the entire ESX server and that VSA node if a single HDD failed. You can do that without losing data if it was all 2-way replicated but it takes time. I contend disks fail more often then many people think they do. Most SAN/iQ hardware/native platforms run internal RAID, they all default to RAID5. If those were my DL380s and I was doing a clean instal I'd make one large R5 set out of all disks, install ESX, create one large datastore, and then install a VSA that uses almost all the datastore available. I'd leave about 10 GB or more free though just in case you need a place to copy something on the ESX server. R1 would work too, just leaves you with less usable space and once combined with 2 way replication means higher redundancy than you might need.
Adam C
Product Manager
LeftHand Networks
|
|
|
|
|
Forum Newbie
Group: Forum Members
Last Login: Thursday, October 25, 2007 3:18 PM
Posts: 2,
Visits: 5
|
|
Thanks Adam. I don't know what I was thinking when it came to the VSA. The only problem would be if both DL380s went down at the same time. Makes perfect sense.
I probably replace 10 or 12 disk a year with my customers so I hear you on the disk failure rate.
I am feeling really dumb but can you point me to the VSA manual? I have searched high and low on the partner web site and all I can find is SAN IQ manuals.
|
|
|
|
|
Supreme Being
Group: Moderators
Last Login: Yesterday @ 3:37 PM
Posts: 108,
Visits: 375
|
|
| Sorry, I'm the dumb one actually, the manual and a new quick start guide are in the version that should go live very soon. If forgot about the lag between creating it and actually releasing it. BTW that version will install those 2 docs along with the UI. They'll get posted separately on the resource center too. If you are dying to read it PM me somewhere to send it to.
Adam C
Product Manager
LeftHand Networks
|
|
|
|