Month: February 2015

Exchange 2013 Preferred Architecture

Here’s a cut-to-the-chase summary of The Preferred Architecture blog by Ross Smith IV, you will find a link for the original article at the end of this blog.

Here’s the summary (with el touch beta3y):

  • Simplicity, Simplicity, Simplicity.
  • For each exchange service (OWA, SMTP,…) Use one name that balances the connections between the 2 datacenters (If you have 2 Datacenter with “fast” network connection).
  • Each datacenter is a separate AD site.
  • All Exchange servers are multi-role servers (Each Server have Mailbox and CAS roles).
  • All Exchange servers are Physical servers!!!
  • 2 Disks in a RAID 1 to host the OS, Exchange binaries, protocol/client logs, and transport database.
  • JBOD SAS 7.2K disks with large capacity for Databases files and logs.
  • AutoReseed is enabled, and at least 1 disk is reserved for it.
  • DAG is stretched across the 2 Datacenters, and active copies distributed equally across all servers in the DAG.
  • Each Datacenter have the same number of Servers in the DAG.
  • The DAG have even numbers of servers, and a witness server is used for for quorum arbitration.
  • DAG’s witness server is placed in a third reliable Datacenter.
  • Use a single network card that will carry both MAPI and Replication traffic (Assuming you network infrastructure can provides 10 Gb end-to-end).
  • Each database has four copies, with two copies in each datacenter
  • ReplayLagTime is 7 days for the lagged database copy (Set-MailboxDatabaseCopy -Identity DatabaseName\MailboxServerName -ReplayLagTime 7.0:0:0)
  • Backup, Backup, backup.

References:

The Preferred Architecture

Configure AutoReseed for a database availability group