Need backup guidance for on-prem and cloud

[jgq85]

Hello,
I was wondering what advice and/or solutions you have for the situation I'm in.
I am trying to understand the logistics of attaining a backup solution where I backup on-prem, then backup the on-prem to the cloud.

Currently we are all virtual servers with VMWare.

We use Veeam to backup VMWare to a Data Domain. The Data Domain goes a bit slow for recovery, so I'm wondering if it'd work to put another backup appliance that we back up to first, which has just the last 1 - 2 days of retention, in order to more quickly spin up VM's in a recovery scenario or file level recovery scenario.
So if we had that, then we continue to keep the past 30 day retention on the data domain.

Since that's all still on-prem, what method would you use to then backup the on-prem backup repositories to something like AWS? Would the Data Domain be configured to backup there? Or would you tell Veeam that after it's done backing up to the Data Domain, to then backup to AWS. Would that be another backup of the VM or a backup of the Data Domain?

[csinetops]

If you have the budget I would add some higher tier storage as your primary backup target to satisfy faster backup and recovery scenarios. Then do backup copies and let your Data Domain handle long term retention.

We run our backups to higher tier storage in our data center with 5 day retention. We then run backup copy jobs to a NetApp AltaVault that is sitting in a IDF on the other side of our building. The AltaVault live streams out to Azure to adhere to the 3-2-1 backup rule as well as long term retention of 7 years.

Not sure if Data Domain has a any features like a AltaVault. You could also use cloud connect and work with a Veeam partner to get your data off site. We use cloud connect for one of our branch offices that has no IT staff to swap tapes or anything like that. Rather than pushing all that backup data across our PTP we go direct from production backup at the branch to the cloud VIA a backup copy job. Pretty cost effective and has worked well so far.

[jgq85]

We then run backup copy jobs to a NetApp AltaVault that is sitting in a IDF on the other side of our building. The AltaVault live streams out to Azure to adhere to the 3-2-1 backup rule as well as long term retention of 7 years.

Yes, I think that's what I am looking to do in my environment (assuming I can get budget between Nimble or Dell/EMC, for about 20 TB to be 5 days retention.

Also, what do you mean by "backup copy jobs" from high tier storage to NetApp? Is that a script or something? Does Veeam execute that backup copy job or is that on the high tier storage or the NetApp to handle that job?

Would I lose DDBoost on the Data Domain if I'm using a backup copy job in a script?

[foggy]

Backup copy jobs are specifically designed for archiving backups to a secondary storage and can be targeted to DDBoost repository just like regular backup jobs.

[csinetops]

Yes, I think that's what I am looking to do in my environment (assuming I can get budget between Nimble or Dell/EMC, for about 20 TB to be 5 days retention.

Also, what do you mean by "backup copy jobs" from high tier storage to NetApp? Is that a script or something? Does Veeam execute that backup copy job or is that on the high tier storage or the NetApp to handle that job?

Would I lose DDBoost on the Data Domain if I'm using a backup copy job in a script?

My backups look like this:

1) Nightly Veeam jobs are run and stored on my faster backup storage (veeam repositories) , retention of 5 days.
2) Nightly Veeam backup copy jobs are run to pull that data from my main backup storage to the AltaVault appliance( also veeam repositories setup as CIFS shares in Veeam) . Retention of 30 points plus 84 months to meet company rules.
3) The AltaVault is a NetApp appliance that can live dedupe, compress and stream whatever data you put onto it out to a host of cloud providers. In my case I use Azure storage to push my backup copies to for off site copies as well as long term storage. The data is still kept on the AltaVault on site for a few months but is then purged based off rules you set on the appliance, then only remains in the cloud. Veeam doesn't know about this as it thinks all the backups are still local on the AltaVault. If you need to restore and its not on the appliance, its pulled back down from the cloud.

I am a 100% NetApp shop so you will probably want to contact EMC during this project to see if they offer anything like this. Even of they don't, give the AltaVault a look, its pretty slick for long term retention both on and off site.

[jgq85]

Goot it. So as far as the backup copy job, that can be configured in Veeam to run after the the first initial faster backup job is completed?
Or do you have to just guess when it will be done by and schedule it around a time after the initial backup job is done?

Out of curiosity, is Azure pricing on par with AWS? We're also a small university with free Office 365, so I'm wondering if there is any incentive to stick with Azure if there are any discounts.

[csinetops]

In your main Veeam job there is a section for a secondary location. If you put your backup copy job info here it will sync as soon as the main job is done, it basically looks for new restore points in the main job and kicks off the copy job once it sees new points.

We are a Microsoft E4 customer so I'm not 100% sure on pricing vs AWS. I can tell you that at our pricing, 6TB of storage in Azure costs something like ~$150 a month. The AltaVault is deduping and compressing our data ~11x before sending it to Azure, that's where the device pays for its self.

Ask your EMC rep about EMC Cloud Boost, that seems to be their equivalent to NetApp's AltaVault.

[jgq85]

For my first tier storage for the quick backup & recovery (with a couple days retention), since that is straight up storage and not a archive solution, then does the total amount of storage on that going to need to match the size of all the VM's I am backing up?
If I have 20TB of production VM, then my first tier storage backup repository will need to be at least 20TB (given it will have a couple days worth of retention)?

[foggy]

Backups will actually take less, considering compression (which is 50% on average). You will need space for the full backup and a couple of increments (only changed data), however, space for an additional full would not hurt, in case you'd need to perform compact, for example.

[jgq85]

Ok thanks so is that Veeam doing the compression?
Or should I look into a particular backup device for the fast backup & recovery that does good compression but is not as slow as Data Domain?
I have Nimble storage in mind, but my boss leans more toward Dell/EMC.

[foggy]

You can configure compression settings while creating the job.

[JJG]

Is there any benefit or cost savings with using Cloud Connect vs NetApp Altavault? My customer is using a NetApp system as the primary target for VM backups and I am comparing options between using Altavault and Cloud connect. NetApp has a Altavault cloud appliance that could be stood up in Azure to recover data in the cloud. So, at the secondary site, there would be no need to have physical presence there to replace drives, etc. In addition, NetApp now supports sending data directly from its storage systems (i.e. FAS8200) to the Altavault, which seamlessly ships to the cloud via SnapCenter.

[csinetops]

I don't see any reason to use cloud connect over a AltaVault. We use cloud connect for one of our remote sites. It works well in our scenario but is much more expensive that Azure and you don't get any of the features you outlined above. It would cost us $1500 a month to store our data in cloud connect (13.6TB now) while Azure is $180. IMO They are two kind of different use cases though. Altavault is meant for some local copies with long term retention in the cloud. Cloud connect would work for long term retention also but would be extremely expensive.

[csinetops]

Ok thanks so is that Veeam doing the compression?
Or should I look into a particular backup device for the fast backup & recovery that does good compression but is not as slow as Data Domain?
I have Nimble storage in mind, but my boss leans more toward Dell/EMC.

Veeam will do the dedupe and compression on the main jobs. Not sure about EMC but NetApp tells you to turn those settings off in the copy job and let the AltaVault handle it.

Whatever you go with, you'll need faster primary backup storage for your backup jobs. Then a backup copy job will copy to your secondary storage/dedupe appliance . From there to the cloud. I personally would never go directly to a DD or any dedupe device, you'll have slow restores no matter what because you have to re-hydrate all the data.

Typically with data sitting in my main repositories I can have a full VM back up and running in <15 minutes, FLR in <3 minutes. From my AltaVault, on prem <7-8 minutes, cloud depends on how big the job is but typically the same day.