Loading times veeam Explorer

[Bjoern_ch]

Hi all,

since we upgraded to v8 a year ago we have problems with the loading times of SharePoint and Teams Explorer for larger organisations (+30K sites/teams).
It still (after some improvements) takes up to 15 minutes that the explorer finishes loading the orga. After that the restore can be performed. Restore itself performs normal.
After working on it for nearly a year, support just told me that this is "not abnormal".

I wonder if other people have the same issues? Do you have to wait 15 minutes (or more) until you can start with the restore too?
What are your loading times?

Best regards,
Bjoern

Case #07798810

[AlexL]

Yeah, got the same, anything over 10k is very slow unfortionately. I've timed 30 to 45 min on occasion. Customer using the self service portal as a restore operator is reporting the same.
Only happens for onedrive, sharepoint and teams, exchange is ok.
What I see is that compute (cpu/mem) is very low, only cpu is consumed by postgresql, and there is some disk read activity for the backup365 database but not very high.
My guess is that some serious optimization needs to be done.

Case #08134270

[DaStivi]

Hi, I've opened a case on this issue and its in RnD already.. they are working on better multiprocessing and optimizations on this issue!

opening/connecting, a big customer took around 30 to 45 minutes for me... expanding the org took another 60 minutes!

one thing i noticed with support together, even smaller tenants are affected! you can compare exchange online vs. onedrive/teams or sharepoint explorer... EXO always opens quite fast... the later 3 not...

when you open the restore through a "job" and NOT through the org-level, the onedrive/teams/sharepoint explorer also opens a lot faster!

one other thing I did in the meantime, i moved from a dedicated Windows postgresql server to a Linux Postgresql, my internal tests showed now opening/connecting the onedrive explorer is now down to a couple of minutes, and expanding the Org is at 15minutes! so PSQL performance is very important on this. i noticed hight load on the windows psql server while opening the restore explorer...

[AlexL]

How many accounts roughly for the tenant and how much cores/memory for the windows server, just to compare.
Could you also share your case #

[DaStivi]

like 13-15k Users, 16 User jobs, without around 1900-2000 Objects each.
Sharepoint/Teams job with 13300 Objects
overall 900TB Data

but as said also smaller orgs are affected too.

my case #08089250

the windows psql has 32GB and 8CPUs, but its more of a Disk IO thing... in my case all the VMs are on NVME-SSD so couldn't be better i guess...
VeeamBackup365 DB is around 290GB, and i've around 70 customers on this environment

[Bjoern_ch]

oh...
on one side it feels good not be alone but on the other side your feedback is worse than I was hoping for
My case is with RnD for 9 months! And it feels like it will be there for another 9 months...
We observed the same with smaller customers, exchange is noticeably faster than SP/Teams.
First, our loading times were even +60 minutes for the large orgas. After adding parallel processing for the Explorers via proxy.xml it became better to 5-15 minutes. But with v7 it was for all customers less than 2 minutes.
Postgres utilisation is quite low with us as well. Our postgres has 48 cores, 96GB RAM and SSDs with more than 5000 IOPS (veeam tested 16000 IOPS). If one single restore job is too much for the db server (we test when no other job is running), well, then it is not the server which has the problem here I guess.

Yes, job based restore is somewhat ok. But in a Service provider Scenario the customer has only one option which is orga based restore.

[Bjoern_ch]

Do you have PGBouncer in use?