v12 SOBR backup to copy job conversion

[mcz]

Hi guys,

currently we're having a BCJ that copies to a SOBR and from there it get's copied/offloaded to the cloud-objectstorage. Now in v12 you can directly copy to the object storage but once again there's the question about the existing jobs: Can you use an existing chain in one bucket and then continue with that chain with your new BCJ (in v12)?

I fear that we will be told 'you cannot map the existing backups to your newly created BCJ, it will create a new chain in your bucket' but then we loose again all the deduplication we had before. This is always the issue with multiple 'new starts' we had before, if you create a new job (for the same source-vm's) you loose all the deduplication.

If this was not implemented in v12, will it probably get shipped later?

Thanks for letting us know!

[HannesK]

Hello,
it's to early to say which kind of mappings will work. There are many "under the hood" changes in the core engine in V12. We also work on improving the mapping between different jobs. But as I said, it's too early to say which things will make it to the GA version.

Thanks for your understanding,
Hannes

[mcz]

Sounds good, thanks Hannes

[Gostev]

One thing to keep in mind is that "Direct to Object" feature uses the new format of storing data in object storage, which is significantly different. This change was needed because for starters, there's no longer any local (on-prem) metadata cache available, as is the case with SOBR where it sits conveniently in the Performance Tier. The new format had to be designed to work well in scenarios where object storage is all you have, which in turn raises many performance and cost efficiency questions. Oh well, I guess I don't need to explain why not having a local cache/index sucks when dealing with a storage located on the Internet far far away and while we managed to make this work well, it required significant format change.

[mcz]

Thanks for the details, Anton. I've heard it before that the local cache would be eliminated in v12, but reading your lines here gives me a bit a different picture. Does it mean that you won't have the local index at all in v12 or does it depend if you had an "old" deployment before? When you say that the format changes, does it mean that it also changes the format for all the new increments as soon as you have installed v12 or is the new format only applied on a fresh bucket/chain?

Thanks for the clarifications!

[Gostev]

Well, I know that the original plan was to upgrade "old" backups to the new format. But it's too early to say what the final state will be, as it will depend on the testing results and whether they can make upgrade work reliably in time.