Slow Migration Speed - a lot of possible "Timeouts"

Hello everyone,

we are migrating about 5000 Mailboxes from Domino to Exchange. Our first Try was to migrate directly from Domino (9) to Exchange Online. The throughput was under 300MB/hour and we had tried everythig to get it higher (even MS Throttling was set higher). Because we have no chance to migrate all Mailbox Data from the bigger Mailboxes over Night and have a Timeslot till end of March we changed our Strategy and migrate the Mailboxes in the same onPremises Environment and move the Mailboxes from onPrem to Exchange Online. Unfortunately we are getting about the same results (~250MB/hour, some Mailboxes about 700MB/hour), what is way to slow.

Two Years ago we migrated in our "Sister Company" and hat about 800B/hour per Mailbox with the same Migration topology.

We analyzed the Logs and found, that sometimes there are Gaps, where the Migrator seems to do nothing, the CPU goes to zero and nothing happens. The log writes "migrating message" and after a long time "migrated message".

So after a long Wall of Text, here are my Question: What does the Quest Migratotor exactly do between these two Steps and where could be the bottleneck? Is the Thread waiting for something? Is it writing to Disk? Could the Virusscan Engine be a problem?

Every hint could be useful.  Thank you in Advance!

Best regards

Christian

  • Could the Virusscan Engine be a problem?

    Yes! And even if it not the root for this specific problem, it will likely slow down migration significantly.

    even MS Throttling was set higher

    There are many ways Exchange Online limits migration speed. In MNE, correct configuration of Admin Account Pooling helps working around some of these, so do the right settings on PowerShell throttling. Of course lifting MS throttling does also help.

  • Hi Peter, thank you for your answer. This Migration-Night we diabled the Sentinel Software on one Domino, Quest Migrator and Exchange Server and built an Job with only Mailboxes for this three Systems. Maybe it will get faster - or we can exclude Sentinel from our list...

    Any Ideas on the Gaps in the Migration? We analyzed the Logs further and one random Job had a runtime of about 7 Hours and in that 7 Hours about 3 Hours were Times > 5 seconds without any further Log entries nor CPU Utilization. When the Notes document is being converted, why is there no cpu utilization?

      We also opened an official Case (5027358) our Problem.