Replication Job vs. Backup Job

[bunger]

Just curious what everyone is doing as their primary job creation types - Replication jobs or Backup jobs.

I am running 2 ESX server - one production and one "backup." vCenter/Veeam are virtualized along with about 4 other servers on the production box. I am currently running all jobs as Replication jobs purely for speed of recovery - if production box dies, I right-click the replicated machine on the backup box and am back up and running in minutes. I am not using automatic fail-over ( and won't use it as I want manual control ).

Just curious what everyone's thoughts are.

tia,
Bill

[Gostev]

Hi Bill, it is perfectly fine approach to use replication job as the primary job type when you only have a handful of VMs to protect. On the other hand, if you had hundreds of VMs, you would not be able to replicate them all, because this would require too much storage - you would have to more than double your storage.

So for larger environments, typically the rule of thumb is "Backup all VMs, Replicate tier 1 VMs".

There is one potential serious issue with your approach though. It sounds like you are not doing any offsite backups. What is your recovery plan in case if your office gets flooded, and both ESX servers die along with the storage?

By the way, we do not have automated failover anyway, it is always manual. If you meant to say that you are not using Veeam UI to perform failover, and instead just starting replica VM from vSphere client, then this is definitely not recommended, because as soon as you do this, you will loose ability to re-failover to an earlier restore point (in case if latest restore point does not work).

Thanks.

[bunger]

Thanks for the reply and the storage "issue" makes sense.

I do have a quick question regarding the recommended method to fail-over. I understand losing restore points if using vSphere client, but what is the method to "fail back?" I have searched the manual, FAQ, and the forum and can't find anything specific.

Thanks again,
Bill

[jlehtinen]

Hey Bill,

Just thought I'd chip in - I have a very similar setup to yours - 2 ESX hosts, one production, one for emergency failover.

I run Replica jobs to keep up-to-date replicas on the failover ESX host, and I also run Backup jobs to an independent NAS that has two 1 TB harddrives in RAID 1 (for additional redundancy). I keep about 2 weeks of restore points for each virtual machine. Having the NAS as an additional backup destination is a safety net in case something crazy happens and we lose both ESX hosts.

[Gostev]

Hi Bill, actually failback is simply not present in v5 (so you have to create reversed replication job to move the VM back), but we added failback to v6 (Q4 2011).

[msidnam]

Gostev,
Are you able to share with us how this will work? I look forward to this feature since we will have a DR environment in a colo (one SAN and 2-3 esxi hosts) and we would like to be able to failover to the DR site, run it for as long as we need (ie, hurricane) then failback to production like nothing ever happened. this is great news.

[Vitaliy S.]

Hi Mike,

I'm afraid we cannot share all the details yet, as we keep implementation details secret until the product goes GA for competitive reasons. Thanks!

[msidnam]

Understood. Thank you