Archive Tier and backup Size

[inferno66]

Hello,

We're planning to use Archive Tier (already using SOBR with S3 capacity tier), to replace or AWS Glacier VTL (with tape gateway), as soon as the V11a will be release (because we need to be able to communicate with the proxy appliance with private IP no public.

But I've question regarding storage usage on Archive Tier (S3 Glacier/DeepArchive) .
Will it be the same than storage usage than on Capacity tier?

Actually, for example, we backup 3 jobs (about 9Tb each) on the SOBR and we have 4 Full Backup of each on the capacity tier which consummes only about 27Tb on the Bucket (because it'll store only the first full and the diff blocks if I'm not mistaken)

This will be the same exact mecanism when offloading to Glacier?
If we store each GFS Monthly backup for 10 years for example on the Archive Tier, it'll consumme ony the first full + block diff for 10 years?
If yes than it'll cost so much less than VTL to Glacier

Also on the SOBR, at the Archive Tier there's an option "Store archived backups as standalone full".
It meean that it'll store each backup as a complete backup with its real full size (so no block dedup)?

Thanks and regards

[HannesK]

Hello,
archive tier and capacity tier work a little different, but also in archive tier there is sharing of objects.

It looks like this




Month 1: ABCDE (all 5 new objects are uploaded to Archive Tier, because this is the initial full)
Month 2: XBCYZ (3 new objects are uploaded to Archive Tier: X, Y, Z; but B and C are reused from the previous Month 1)
Month 3: ABCYE (2 new objects are uploaded to Archive Tier: A and E. Even if Month 1 already has both, because we only reuse objects from one restore point before. B and C are still reused because they are available in the previous Month 2)
Month 4: AB12E (2 new objects uploaded to Archive Tier: 1 and 2. Objects A, B and E are re-used

Best regards,
Hannes

[Gostev]

In other words, archiving is still forever-incremental, but delta is from the immediate previous restore point only.

Since archiving is done in a very large blocks, the chance that an extinct mega-block reappears after a few weeks/months is near zero. While not having to looking up all restore points saves on API costs and improves performance.

[inferno66]

Hello

Ok, not simple but I understand the concept.
So if I understand Ii could save some space but has this is mega-block on archive tier, there a big probability that not so many blocks are reused, so the space gain (and cost) is not so huge.

And regarding "Store archived backups as standalone full".
It meean that it'll store each backup as a complete backup with its real full size (so no block dedup)?

Regards

[HannesK]

correct, standalone full creates full backups without deduplication

[Gostev]

There should still be significant savings between weekly GFS for example, but much less between monthly (if you only do monthly) simply because lots of data changes in 30 days. Although even there, static workloads like image/video archives will show excellent space savings.

Unfortunately, Glacier comes with very high API costs which is what dictates the usage of mega-blocks for this object storage tier to make economic sense.

[inferno66]

Thanks a lot for your feeback

Our biggest backuped file server is a media data server so a lot of static content, in this case it should save many space ^^

[HannesK]

maybe you can post in half a year or so how much you saved. That would help other customers

[inferno66]

Yes of course