Feature Request: SOBR load control of concurrent tasks

[edison5000]

Currently SOBR's do not load balance concurrent tasks for each extent. Even after you join new repos to a SOBR, you must still guess how many tasks to set for each individual repository. I suggest that when a SOBR is created, it should consolidate task control for all its connected extents. That way you would only adjust it in a single location.

[PetrM]

Hello,

Technically, it's possible to use different hardware for different extents. Let's say I have 2 extents in SOBR: the first one is able to handle 2 tasks maximum and the second one has more CPU and RAM so it can process 20 tasks in parallel. If I had a consolidated task control at SOBR level, I'd probably encounter a lot of performance-related issues with my first extent. Also, you may have a look at the recommendations provided on this page to find out how many tasks you should set for each individual repository.

Thanks!

[edison5000]

Petr - yes that's true. In fact I thought that was ideally what one was supposed to do. : ) But why not just have fields to plug in specs if Veeam can't detect hardware specs on the fly? I'm suggesting a method for telling Veeam "these are my available resources, use them accordingly, automatically. I know this is difficult, because you could have maybe a 10 year old server with 24 cores that don't perform the same as a brand new server with 24 cores. Nevertheless this is what I am suggesting.


tangentially (already have case open for this): Currently we are using a SOBR with two local repos on different disk arrays, but they are both on same server and raid controller. It feels like it should not need the adjusting that is currently necessary, where we have to keep tweaking the tasks to keep it stable. We chose the Performance tier, but perhaps that was not a smart idea, as it just throws the majority of incrementals and tasks to the new repo that has more empty space, but keeping fulls on the first repo. We are wondering what will happen once they are both equal in capacity; i suspect Veeam is not going to change its behavior when that happens. Either way I have not read anything about SOBR load balancing improvements of this kind for v12.

[PetrM]

Hello,

these are my available resources, use them accordingly, automatically.

I've got the idea but and from my point of view it will make sense only if it's fully automated, there is no point to enter specs with hardware resources manually as it would be the same approach as what we have today: I set concurrent tasks depending on available resources on every extent. Anyway, the request is noted but no ETA as usual, at first we need to correlate potential value of this improvement and required engineering efforts.

I suggest to read this section of our help center to get a better idea of both file placement policies: Performance and Data locality. Basically, extent selection logic depends on the selected policy. In Performance mode, Veeam always stores full and incremental backups of the same backup chain on different extents.

Thanks!