Maximum Bandwidth limitations for Veeam Backup for O365

[Gseehomie]

Hi There

We'd like to calculate the estimated time for a customer on how long it will take him to do the initial backup and after the inital backup how long the incrementals will take.

Last time we tried it we had roughly a bandwidth of 20 Mbps, although our company is capable of way more than that. Now my question for the community is if O365 has a bandwidth limitation in general or for Veeam O365 Backup? What could be the cause, that the bandwidth in our test was limited something around 20Mbps?

Best regards and many thanks in advance!
Tizian

[Mike Resseler]

Hi Tizian,

First: Welcome to our forums!
Second: We don't have a bandwidth limitation but O365 is limited at a certain point in time. When you are downloading too many objects at the same time, you can get throttling. Those throttling rules are set by Microsoft (and through their support you can get some adjustments but not unlimited). There could be other reasons also. Memory/ cpu of the VBO or Proxy server for instance.

[Gseehomie]

Hi Mike,

Thanks for having me here!

Alright, I assumed, that Veeam is not limiting the bandwith of the connection to O365. If someone is throttling the bandwidth it had to be Microsoft. Memory and CPU were enought at the time beeing. But you don't have any concrete numbers with which we could work with? I don't feel conftable telling the customer that we can't calculate the time he needs for a certain amount of data to backup. Does Veeam have any advice in this case or do i have to contact MS-Support for further details?

Best regards and many thanks!
Tizian

[Mike Resseler]

Hi Tizian,

Unfortunately I can't help you here. The maximum we see is over a TB per 8 to 9 hours, but that is in ideal circumstances. We have logic implemented, allowing us to avoid the backup being throttled, and that makes us faster as others, but it will only work if you have multiple mailboxes/ drives/ ... in one job. And if the last (mailbox for example) is a very large one, and all the others are done already, then that part will seriously slow down.

But since you need to say something against the customer, I would advise you to say around a TB per 12 hours in a good setup should be doable

[Gseehomie]

Hi Mike

First of all i'd like to apologize for the long delay in my answer. I was on holidays for 3 weeks and didnt check the veeam forum.

Thank you very much for the answer stated above! I've contacted the customer and now it seems that the problems shifted a bit from bandwidth limitations to scanning time. They have the following szenario:

The customer has around 320 objects to backup every day. Now every day veeam seems to scan every object for changes and then synchronizes the changes. For the scan veeam needs around 5 hours. To synchronize the changes they just need a couple minutes after that.

1. Is it normal, that the scan for all the objects does take that long (it looks like especially Sharepoint sites and personal sites take a couple minutes each), even if nothing changed during the past day?
2. How is this handled with big customers with x 1000 objects to backup? I assume the scan would need more than 24 hours and the backup cycle is every 24 hours.

I really appreciate your help and many thanks in advance!
Tizian

[Polina]

Hi Tizian,

Mike will correct me if I'm wrong, but 5 hours seem to be a quite possible scenario. It does take time because items are checked one by one and their versions are compared to those in the repository. The number and size of items also matter - a job with N GB of bigger items will run faster, than a job processing the same N GB of smaller items.
You can fine-tune the performance by adjusting the number of threads and/or deploy additional proxies to distribute the load, but still, the versions check is rather time-consuming.

[Mike Resseler]

Tizian,

I do suggest to create a support call. 5 hours before it starts to synchronize the changes doesn't seem too normal if we are talking about 320 objects. As Polina says, it can depend. (One object can be 1000 of files but still...)

Our support engineers should be able to read more in the logs, although it could even be that extended logging needs to be enabled. It can just be 1 object that is causing this.

Please post the case ID and outcome after investigation with the engineers here

[Gseehomie]

Tizian,

I do suggest to create a support call. 5 hours before it starts to synchronize the changes doesn't seem too normal if we are talking about 320 objects. As Polina says, it can depend. (One object can be 1000 of files but still...)

Our support engineers should be able to read more in the logs, although it could even be that extended logging needs to be enabled. It can just be 1 object that is causing this.

Please post the case ID and outcome after investigation with the engineers here

Hi There

It seems that the customer found a solution for the problem stated in my previous posts. They created 2 different backupjobs. 1 for Sharepoint backup and 1 for OneDrive Backup. In addition they created a powershellscript which allways adds new elements to the existing backup job (sadly i don't have access to that script -> would be very interesting for me to see how this works). Now the backup takes around 40 minutes which is a huge improvement in comparison to the time needed before (do you have any reasonable explenation for this behavior?)

Looks like this will work for them.

Best regards
Tizian

[Mike Resseler]

No. It doesn't make sense to me that the backup is now much faster just because they started to use a PS script to add new users or sites.