Veeam job question, year's worth of data.

[Lanz]

Moving from commvault to veeam.

Have 2 physical backup boxes acting as proxies and repositories, one at each site. Doing network based backups at the moment.

If we wanted to keep a years worth data that is stored both locally for quick restores and remotely for DR whats the best way of doing this.

Right now with testing, we are running 2 jobs for each group of VM's, a local reverse incremental job every night, with 365 restore points, with a full backup once a month. And straight after that job is finished, we run a forward incremental job with a synthetic full on a sat with 365 points that gets sent to the remote repository.

Seems like a crazy way of doing things to me, running two jobs doing the same thing, but it seems there's no decent way of copying to a remote site.

Do we make single jobs with 365 restore points, or do we create multiple, smaller restore point jobs for weekly, monthly etc?

thanks

 

[wasc]

You're right - each backup job targets only one repository, so for a local copy and a remote copy, you'll need two jobs.

What storage are you using? For example you'll need different settings if you're using Windows 2012 and it's de-duplication technology vs another storage (eg windows 2008 server share) on your repository.
To be honest, with a years retention as a requirement, I would suggest some form of deduplicating appliance, Windows 2012 being one way of doing it. There are others (e.g. Starwind iscsi software San), but 2012 is probably the cheapest way of doing it with some level of support.)

We are using Windows 2012 with dedupe as our repository, and Veeam is configured to do a daily incremental with a weekend synthetic full, and a monthly active full.

Do you want your "years" retention stored both on-site and offsite?

 

[Lanz]

We are using 2 win2008R2 servers, Dell R720XD with 34TB for the repository (RAID6)on each.

Some people seem to do forward incrementals and then just rsync/robocopy the files over to the other site rather than running 2 jobs, but with that you cant really use reverse incrementals as all files get touched.

We really need to store a year at both sites, just incase one repository gets wiped.

So if we used Win2012, would a year's worth of those weekly/monthly full's get deduped by the OS? I've not looked at 2012, we dont have any currently on our domain.

Obviously only some of our VM's need to be kept a year (user data), the majority of the app servers etc just need 3 months. We have about 200+ VM's with a 70/30 split at each site.

 

[DRZ]

Switch from reverse incrementals to forward incrementals, enable synthetic fulls and you're more or less in the same boat you're in now. Then scoop up the backups and send them to your offsite location.

That's how we do it, only we offsite onto LTO6 instead of another hot backup (just in case). Our servers run 2008R2 but the data is stored on a NetApp which dedupes the data.

 

[Lanz]

Thanks for teh info, with forward incrementals, can we just use standard robocopy transfers, or does it need a block level, rsync type copy?

Also, what's the impact of restore times when using forward rather than reverse, and can you use vpower instant recovery still?

 

[DRZ]

They are slower restores as you get further from your last full (as you would expect).

We scoop ours up with an incremental job (weekly full after the weekend active fulls have run) so I'd go out on a limb and say that robocopy would be able to manage it just fine with the right flags. No harm in setting up a new repo on some junk storage somewhere and running a quick test to prove it out.