Having issues during mailbox moves in Exchange 2013? 

  • Article

Few quick ways to check your mailbox move performance 

 

What are the major factors involved/impacted during a mailbox move?

 

  • Disk IO
  • Network Bandwidth
  • Size of the mailboxes moved
  • No. of active connections on the source and the destination databases
  • MRS throttling

 

Ideally every individual who is troubleshooting a mailbox moves will have the initial probing done and that has to be in place before proposing a workaround or fix.

 

If Disk I/Os are not optimal, we might have severe effects on the mailbox moves resulting in read/write latency. Ensure that the disk I/O read/write speed is optimal (0.020 ms is considered optimal), Also ensure the disk drivers are up to date.

 

In order to get a quick insight about your mailbox move performance make use of the mailbox migration performance analysis script. This script provides you with all the necessary information,

 

Mailbox migration Performance analysis script

 

  1. It provides the amount of data transferred (in GB)
  2. Duration of the migration batch & also the idleness info
  3. Best part is that, this also reports latency which will help us take a proactive call to suspend the moves before it hit the overall disk/server performance.
  4. Also its easy for reporting and stuff

 

Script used to extract the above statistics and the procedure is explained in the below TechNet Blog,

 

https://blogs.technet.microsoft.com/exchange/2014/03/24/mailbox-migration-performance-analysis/

 

Good to know information,

 

Also in addition to this, in order to gauge the load on our exchange 2013 servers we need to get the number of active connections (OWA, ActiveSync, RPC). I have not covered EWS connections in this blog,

 

Get-Counter "\MSExchange OWA\Current Unique Users" -ComputerName “Servername1, Servername2”

 

Get-Counter "\MSExchange RpcClientAccess\User Count" -ComputerName “Servername1, Servername2”

 

Get-Counter "\MSExchange ActiveSync\Current Requests" -ComputerName “Servername1, Servername2”

 

Stages of mailbox moves:

 

Source enumeration  > Initial data sync > Incremental Sync > Finalization

 

I will post a detailed blog on each stage of the mailbox moves in near future :)

 

Other possible fixes:

 

  1. Modify the config file – MSExchangeMailboxReplication.exe.config,

            ‘C:\ProgramFiles\Microsoft\Exchange Server\V15\Bin’

 

 

            MaxActiveMovesPerSourceMDB=”20″

            MaxActiveMovesPerTargetMDB=”20″

            MaxActiveMovesPerSourceServer=”100″

            MaxActiveMovesPerTargetServer=”100″

            MaxTotalRequestsPerMRS=”100″

            ExportBufferSizeKB=”512″

 

            Also can increase MaxConcurrentMigrations and set that to 100 for a fair mailbox move performance

 

  1. Increase the ExportBufferSizeOverrideKB

 

            This way we are adding moving more data at a time which is best supported in a network which has

            high latency networks >100ms

 

            This reduces the initial data sync duration which covers the 70% of the move process

 

            Note: Buffer size cannot go beyond 10240. but its recommended to apply this stage by state and the value has to be tested gradually.Also there are high chances of network chocking.

 

  1. Disable indexing on the source & Destination databases

 

            Personally I wouldn’t recommend to do this, as this has other adverse effects :)

 

 

This is not it, we have a lot more to look at based on the nature of the issue. So drill it down to the core to find the best solution.

 

Regards,

Ganesh G