Backups, replication, both or?

[ashman70]

Currently I am doing vm backups in a Vmware environment to a local NAS. I am upgrading hosts and want to reuse the old host for backups but I am not sure which strategy to go with. The client is in the health care field, their main vm runs practice management software and once its on the new host I want to reconfigure their veeam job to backup every 15 minutes. The old host is a Dell T610 with 24GB of RAM and 500GB near line SAS drives attached to a perc 6i in RAID 5, I want to replace the drives, likely with four 600GB 10K SAS drives in either RAID 5 or RAID 10, I could put Esxi on it and use it for replication, or I could put Windows Server 2008R2 on it and use it as my Veeam repository for the main vm, the idea being if I ever had to start the vm from a backup it would be on fast drives, the new VM host has 900GB 10K drives in RAID 10, but no vm's have been migrated to it yet.

My goal is to be able to recover from a disaster, either at the vm or host level, quickly and easily with the minimal loss of data in the shortest possible time.

Would replication be better then backups? If I have one VM host and one Veeam backup license, can I install Veeam on two servers and have two repository's backing up different vm's from the same host without requiring a second license?

[Vitaliy S.]

I would recommend to do both, backups for long-term archival purposes (I know that health care companies usually has such requirements) and run replication jobs with weekly retention policy for quick failover when the DR situation happens.

As to installing two backup servers, then you can do that, assuming you will be protecting the same source host, but not sure I understand why you can't do this with only one server?

[ashman70]

Thanks, but if I only run replication jobs weekly, how does that help, the data would be a week old? My idea about two backup servers is that one of them has slower near line SAS drives, the other would have faster 10k SAS drives, if I needed to start the main vm from a backup on the server with faster drives, even as a temporary measure, the performance would be better on faster drives.

[foggy]

Thanks, but if I only run replication jobs weekly, how does that help, the data would be a week old?

Vitaliy is talking about weekly retention (weekly worth of restore points), not weekly replication cycle.

My idea about two backup servers is that one of them has slower near line SAS drives, the other would have faster 10k SAS drives, if I needed to start the main vm from a backup on the server with faster drives, even as a temporary measure, the performance would be better on faster drives.

You can achieve the same by configuring two backup repositories within a single Veeam B&R console.

[Vitaliy S.]

Got it, but you can just create two repositories located on fast and lower end storage and avoid managing two backup servers.

As to replication job question, then I would suggest replicating mission critical servers more often, for example every hour (or less) and use retention policy for 1 week, and use backup jobs to save data on the repositories on daily basis (not every 15 min).

[ashman70]

So you would recommend replication jobs more frequently then backup jobs? Any particular reason why?

I understand now the idea of one backup server with multiple repositories, its so obvious I don't know why I didn't think of it.

[Vitaliy S.]

Yes, of course. In case of DR situation you can quickly start the required VM using Veeam backup console or from the vSphere Client and not wait for full VM recovery process. Backups are good if you need to do FLR operations or save files for archival purposes, and backups are stored in compressed and deduped format as opposed to VM replicas.

[ashman70]

Is there any reason not do to replication jobs say every 20 min? I want the minimal window for data loss possible.

[Vitaliy S.]

If your infrastructure can support snapshot commit operation of the VM with no negative impact (see this topic for more info), I don't see any reasons why you cannot do this. Thanks!

[ashman70]

Are you assuming my environment can support snapshot commit with no negative impact, or asking? Should I upgrade to V7 to achieve better performance and avoid snapshot commit issues? Does replication handle snapshots differently from backup jobs?

[Vitaliy S.]

Backup and replication jobs are treating VM snapshots in the same way, but replication jobs provider better full VM recovery times at the cost of more costly target, and more disk space used. As regards the upgrade question, then it does make sense to use our latest version for sure.