Host-based backup of VMware vSphere VMs.
Post Reply
JamesMcG
Enthusiast
Posts: 39
Liked: 8 times
Joined: Jul 11, 2012 3:39 pm
Full Name: James McGuinness
Contact:

Replication to DR site options

Post by JamesMcG »

Our company is new to vmware and veeam, we're about 6 months in at HQ and about to start converting prod machines to vmware. We are using Veeam in prod and dev, and all is well. Love the product honestly, BackupExec can suck it :evil:

Anyway, I'm trying to come up with the best replication scenario for us. I'll try to sum it up:
- Prod is 25-30 VMs, 4 hosts, Equal Logic SAN
- DR is completely equal hardware, 1GB LAN-X
- Current setup, our main data is raw mapped SAN vols replicated by the Equal Logics to the DR evey 10 minutes

I have 3 options the way I see it, first:
- No raw mapped volumes, all vmfs
- Veeam replication of all vms to dr

Second
- no raw mapped vols, all vmfs
- Equal Logic replication of all (3) datastores to dr

Third
- raw mapped primary / large data vols
- Equal Logic replication of all datastores and primary data vols

We are counting on the first option, but I ran a 10min rep job of a few very small idle VMs for 1hr and well I don't know if this will work out. Rep job takes more than 10min to run, if we add in the real prod machines I see no way for this to work. Maybe my 10min rotation is a little too aggressive? We have this in contracts with the customers however...

The second option, I'm not sure how vmware at the DR will react to this crash consistent datastore. Can I rely on this method?

Third is not ideal, Veeam backup will not run on raw mapped vols right?

How do you guys handle your DR replications?
Gostev
Chief Product Officer
Posts: 31460
Liked: 6648 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Replication to DR site options

Post by Gostev »

The requirement of 10 mins RPO for *all* VMs does not leave you many choices really... you have to use Equallogic replication to achieve that.
dellock6
VeeaMVP
Posts: 6137
Liked: 1928 times
Joined: Jul 26, 2009 3:39 pm
Full Name: Luca Dell'Oca
Location: Varese, Italy
Contact:

Re: Replication to DR site options

Post by dellock6 »

Hi James, storage replica by default is crash consistent, not application consistent. It replicates every written byte from one storage to the other, but you have no commit on application like databases or other stuff. Crash consistent replica is fine for the vast majority of applications around, but there are situation where you need application consistency, like databases.
I'm not an expert at all on EQL, but I was pretty sure they have plugins and drivers to do application-aware snapshot, since many mid-high end storage can do this. I crawled a bit on internet and found this:

http://www.dellstorage.com/WorkArea/Dow ... px?id=1571

you would probably protect at least your critical VMs with this add-on. Again, I do not know how it works and if it a good product, I think is a small driver you would have to install on the VMs you need to protect, just like in other storage vendors offering this feature.

Be sure also to do your maths carefully, snapshots consume disk space, so correct sizing of the EQL is pivotal.

Luca.
Luca Dell'Oca
Principal EMEA Cloud Architect @ Veeam Software

@dellock6
https://www.virtualtothecore.com/
vExpert 2011 -> 2022
Veeam VMCE #1
Gostev
Chief Product Officer
Posts: 31460
Liked: 6648 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Replication to DR site options

Post by Gostev »

Actually, I would not try to make application consistent replicas of all VMs every 10 minutes... too much impact on the applications this way with all the freeze/thaw activities. When very low RPO is needed, you usually want to go with crash-consistent restore points. You will still have "good" application-consistent backup in rare cases when restore points produced in such way will be unrecoverable.
JamesMcG
Enthusiast
Posts: 39
Liked: 8 times
Joined: Jul 11, 2012 3:39 pm
Full Name: James McGuinness
Contact:

Re: Replication to DR site options

Post by JamesMcG »

Gostev wrote:Actually, I would not try to make application consistent replicas of all VMs every 10 minutes... too much impact on the applications this way with all the freeze/thaw activities. When very low RPO is needed, you usually want to go with crash-consistent restore points. You will still have "good" application-consistent backup in rare cases when restore points produced in such way will be unrecoverable.
I more or less proposed the same solution yesterday to my collegue, so it's nice to hear it coming from you! We keep the prod veeam backups at the DR site anyway, so they'd be 'onsite' in case of a true total loss disaster at the primary.

As Luca was mentioning we have Auto Snapshot Manager available to us from Equal Logic, I just need to explore how I can tie that into replication. In which case our dbs can be aware, and the rest can be crash consistent. We'll then use Veeam replication for a few HQ/Dev machines to towards the prod/dr sites. Should be a strong mixed solution I think.

Thanks for the input guys.
Post Reply

Who is online

Users browsing this forum: KonstantinS and 99 guests