Help to new Backup storage solution

[enoch]

hi everyone,

Going to create a new Veeam Backup & Replication solution and need to get cons/pros

Solution 1)
New HPE Apollo 4200 with 2xCPU, 256 GB Memory, 2x480 GB SSD for OS in RAID-1, 20x12 TB SAS drives in RAID60, 2x12 TB as Hotspace and 2x1,92 TB SSD for HP Smart Array Cache.

Installed with Windows 2016 and creating Backup storage drive with REFS (hopefully stable now)

Solution 2)
DELL/EMC DataDomain 6300 with 80 TB capacity (can be expanded with disk shelf). DELL says this should compare with the 180 TB REFS in solution 1 because of better dedupe / compression vs. block cloning (REFS)

I need to Backup Copy data to two other locations with currently is servers with Windows 2016 REFS with local storage. If solution 2) is the way to go, then we could in the near future buy one more DD6300 and do replication between them.

Price is "almost" the same when comparing the two solutions.

[Andreas Neufert]

Global dedup targets have by design a random IO penalty, so that you will get limited restore performance.

Specifically over time the Apollo will be much faster.

We see storage replication not as a second backup copy as it is the same Veeam chain.

So I would place the Apollo on site 1 and do a Backup Copy Job to a Server (Veeam Gateway) on site 2 that use the Datadomain with DDBoost as long term retention target.

[enoch]

Hi Andreas,

Thanks for reply.

Today we also do Veeam Backup Copy to two other targets. Same reason as you write about - if replication is done native from DD to another DD then "same" possible errors will be replicated.

I also think that doing the Apollo on primary site and then for future Investments one DD box at another site and then maybe some kind of Tape solution on third site.

If using Apollo on primary site and having a DD to do replication by Backup Copy then Veeam Encryption is not possible right?

[csydas]

Hi Peter,

Maybe I'm wrong on this (would love a green poster (hint; @Andreas) to comment), but I would imagine the cross system encryption is no big deal. The DD will see the encrypted blob, see that there is no dedupe opportunity, and write it straight.

For what it's worth, I'm with Andreas; 180 TB on 80 TB is a pretty bold claim, and my experience is that the benefit of ReFS block-cloning is gonna be a lot better than Catalyst's synthetic operations in virtually every metric. Sure, you'll cram a lot of data onto a small appliance, but if it were me, I'd rather have a fast front-storage and something like a Data Domain as the second target.

[Andreas Neufert]

In Veeam, as we create independent chains, you can select encryption by backup chain.
So you can use encryption on primary side and then use DD encryption (without Veeam encryption) on secondary side.

Overall DD would stop deduplicating blocks when you would store encrypted files as they unique.