Regarding Veeam Backup for O365... if a customer has 30TB+ in O365 (consisting of exchange/sharepoint/onedrive) is there any way of seeding this data down to the on-premise instance i.e. the reverse to AWS Snowball or Azure DataBox ) or is it the case of the data trickling over the public internet connection?
If that is the case - is there a way of throttling/scheduling this download so that backup copy doesn't interfere with the production traffic for the customers on premise internet access?
Plus it may take forever before the initial baseline is sync so do we just accept this and set expectation by calculating the data size vs bandwidth + rate of change and frequency of backups?
I don't really know if there is a "seeding solution" like AWS Snowball oder Azure DataBox, but you can definetly throttle the traffic while backing up.
You can adjust the threads and limit the bandwith per proxy. Modifying Backup Proxy Server Properties
VBO doesn't feature any seeding mechanism. The backup time will depend not only on the amount of data to transfer, but also on the number of individual objects to enumerate and process. However, I'm afraid that it's hardly possible to provide you with any reference numbers to estimate for jobs duration. The results depend much on the specifics of your infrastructure.
To optimize the performance you can, first, create separate jobs for different types of objects (mail, OneDrive, SharePoint). It'd also make sense to split up SharePoint sites between several backup jobs.
Please note, that you can schedule the jobs the way you need it, and even set up to terminate a job if it exceeds your allowed backup window.
It's also possible to fine-tune the performance by adjusting the number of threads and setting limits for the bandwidth (if needed), as @Falko suggested.
Hope this makes things at least a bit more clear )
Thank you for the prompt responses and both very useful details indeed.
I thought this might be the case - in the sense it's not like we have an Azure hub we can just plug and and play some storage in to to preseed the data