StoreOnce Catalyst vs. synthetic full backup

[jobhlidal]

Hello.
Please, can somebody answer these questions.

Veeam backup job uses as a target StoreOnce catalyst (per VM backup)
Once a week Synthetic full backup is scheduled.

I found this message regarding Catalyst and Synthetic full
" Veeam will open a new data session to the StoreOnce for each backup file within a backup chain that is being synthesized"
E.g. backup job has 10VM, each VM has 6 version. It means 60 sessions that would be open to StoreOnce.

Is it still valid?
What happen when number of sessions is higher than Catalyst store connection limit?

StoreOnce Catalyst system limitation uses value "Maximum concurrent data and inbound copy jobs".
Does the term "Data session" matches the term "data and inbound jobs"?

Thank you
Jan

[StevenMeier]

I have not noticed that on our StoreOnce when running but we are 6.5 not 6.0

[Andreas Neufert]

Veeam accesses only the number of dependant files from a single chain per session. If you have one full and 6 dependant incrementals, then we read from 7 files.
Depending on the StoreOnce model the number of files you can access at once is limited. We understand this limitation and analyse if the actual job has to wait until other jobs finish.

[Andreas Neufert]

Moved to correct subforum

[foggy]

StoreOnce Catalyst system limitation uses value "Maximum concurrent data and inbound copy jobs".
Does the term "Data session" matches the term "data and inbound jobs"?

No, concurrent tasks and data sessions are different things, each task might require multiple sessions, depending on the number of files in the chain, etc.

Btw, here's a Veeam/Catalyst configuration guide, might help to address some of the questions.

[jobhlidal]

I know the mentioned guide.
I do not want to mess two different terms.
I know there is a term "Max restore points per single backup chain" which is specific for each StoreOnce box (e.g. for StoreOnce VSA it is only 7).
This limit is handled by Veeam and it does not allow to create backup/copy job with more restore point than is a limit.

The second term is "Maximum concurrent backup sessions per node" (or maybe "Maximum concurrent data and inbound copy jobs".),
I suppose that Synthetic backup is performed in parallel for more VMs at once. To create synthetic backup for VM, Veeam must open all restore points (full + all INCR).
And for each restore point there is a need to open one connection to StoreOnce. And number of these connection which StoreOnce can handle at once is also limited on StoreOnce side.
And I suppose that number of these connection is not the same as "Max restore points per single backup chain".

I am looking for the number of these connection in the technical specification of each model of StoreOnce.
Is it term "Maximum concurrent data and inbound copy jobs"?

[foggy]

The backup chain length (max number of restore points in a chain) is actually defined by the number of concurrently opened files (data sessions) the specific StoreOnce model allows ("Maximum concurrent backup sessions per node" in the table on p.37 in that guide). So you're basically speaking about the same limit. Veeam B&R just makes sure this limit is not exceeded by scheduling its activity accordingly.

There's also the number of concurrent tasks assigned to proxy or repository server, which is a Veeam B&R own term/limit.

[jobhlidal]

I do not think so that both terms are about the same.
Max restore points per single backup chain - it is about number of versions in restore chains for one VM (in case of per-VM backup)
It means that for StoreOnce VSA (limit 7)I can create backup chain only with 1 Ful and 6 Incr, no more.

Maximum concurrent backup sessions per node - it is about max number of data sessions which can be opened at once
When I return to the beginning to the statement " Veeam will open a new data session to the StoreOnce for each backup file within a backup chain that is being synthesized" when you perform synthetic full backup.
If I have a backup job with 3 VMs, each have 7 version (1Full and 6 Incr) and I start to do synthetic full backup, then, according to the statement, Veeam needs to open 3x7=21 data session for reading.
And it can do it in parallel, because value 21 is less than "Maximum concurrent backup sessions per node" which is 64 for VSA.
If I have backup jobs with 10VMs, each with 7 version, then I need to open 70 session to create synthetic full backup.
And because VSA limit is 64 it is not possible to do it in parallel and there will be message that Veeam wait for resources.

Do you agree with it?

[foggy]

Yes, overall your understanding regarding data sessions is correct (there's some overhead in data sessions reservation though, since more sessions are reserved than actually required). I'm not saying both terms are the same, they are different numbers, but one of them is derived from another, since the allowed chain length directly depends on the storage connections limit due to the reasons described above (this dependency is shown in the table).

[jobhlidal]

Thank you for clarification.