Upgrade v7 -> v8 Repository Sync Duration

[AlexL]

We have a number of larger customers that are still running on v7 but we are finally starting to upgrade them. The upgrades so far went fairly smoothly.
The upgrade of the object repositories seems to be the part that takes the longest time, the install/upgrade itself is fairly simple and straightforward.

What would be the most determining factor in the repository upgrade time?

Is it number of items (mails/files etc), or the number of user accounts (or objects) or is the size in TB of the data?

For example, how would a customer with 3600 users and 138TB of data compare with a customer with 8000 accounts but only 40TB of data.
The former was done within an hour (2 repositories), the latter will be upgraded soon.
Previously I've upgraded another customer with 1200 users and 12TB of data in 1 repository, that was done within 10 minutes or so.

These three customers are the smallest of our larger MS365 backup customers.

Furthermore, the repo upgrade and following sync seem to be both compute heavy and disk io intensive. Would it be wiser to upgrade one repo at a time or multiple, or even all, at the same time? Considering the goal would be to minimize the total upgrade time, and as such down time, for all repositories.

[Polina]

Hi Alex,

During repository upgrade, we perform data indexing and also transfer persistent cache to PGSQL. Repository upgrade time depends on the size of its persistent cache. Indexing is a background operation and its duration depends on the number of restore points in the repository. For estimates, check out our upgrade guide.
Upgrade of S3 Compatible repositories may take longer compared to AWS or Azure.
In our tests with generated data, on average upgrading a 1 GB cache took about 68 seconds. On real-life data it may take longer.
Indexing of 1K restore points, according to our tests, completes in about 3 minutes.

I'd recommend you contacting our SEs to get assistance wiht planning your upgrades.

Thanks!

[DaStivi]

Hi Alex,
Ii don't have the thread at my hand, but if you look through my posts i made, you will find something about this... I'm in the same boat... had my service provider environment updated some months ago and I had waited quite some time because of many initial "issues" with vb365 v8 ...

repo upgrade and indexing finished within a few days, if not even in 24h for me...
but be sure you've enough IOPS on your Storage available!

one of the "bigger issues" where the size of the psql Server/DB that grow quite big and almost hit 100% a few times... thankfully monitoring and resizing the partition before it ran full didn't add further issues!

I did the upgrade mostly one by one... but the smaller ones, as you've also experienced where done in a matter of minutes... further more the initial upgrade would take some time but afterwards the indexing starts too... so for the bigger ones some jobs ran in parallel on my side...