Slow pulling down in 3.0 for first full

[dwoodward]

We just installed Veeam Backup for Office 365 today - version 3.0 - customer has a "hybrid" deployment. So far it is running very slow pulling down from Microsoft (3.3Mbs on test job of two users - it's running in the KBs on the full) - the customer has 100Mbs internet. I saw we could ask Microsoft to increase the EWS throttling rate. We requested this and are waiting on the change to occur. Has anyone else encountered this and did you request an increase and did it help? My customer has about 10TB with about 2TB email and 8TB public folders - at the current rate this is going to take forever.

[nielsengelen]

The initial backup mostly depends on the network towards Microsoft but changing those values via them will help a lot.

Could you tell a bit more about the setup on the VBO server in terms of CPU/mem/disks/...?

[Mike Resseler]

Another item... Are those slow numbers on the job with two test users? And do they have lots of (small) items inside? Because if that is the case, it means throttling will be almost instantly. And in that case our "system" to avoid throttling (at least as long as possible because you can't avoid it entirely) will not be able to work. Another thing here (with the two test users) is that you probably need to lower the amount of threads because you are doing too much connections at the same time, causing the throttling to happen faster.

[pesos]

Hi Mike, how do we lower the number of threads? I haven't seen such an option and it seems like all the mailboxes try to go at once, resulting in a huge number of connections and a jobs that is running for days (and still only 1/3 of the way done).

[nielsengelen]

This is done under the proxy settings (more info in our user guide).

[pesos]

Thank you! Is there any guidance on a good number? Looks like 64 is the default?

[Polina]

Hi John,

There's no silver bullet in this case - 64 is the default, and it a more a matter of fine-tuning (using some "test" jobs, for example). If you're running several jobs on the same proxy, the threads will be automatically distributed between those jobs.

Our support engineers can assist you with performance optimizations (if you open a support case, please post you case ID here).