0 Replies Latest reply: May 9, 2014 11:21 AM by julez RSS

    My company's Exchange 2013 deployment

    julez Adventurer

      This is a branch from another thread I had started asking about various which performance policies to use for mount points for Exchange when the host drive was a VMDK. https://connect.nimblestorage.com/thread/1587

      I wanted to throw out our configuration in the hopes it would help someone out there.

       

      Here are some helpful links:

      Exchange Guru Install Guide: http://msexchangeguru.com/2014/03/02/e2013sp1-installationupgrade/

      Exchange Guru IP less DAG (Exchange 2013 SP1 on 2012 R2 ONLY): http://msexchangeguru.com/2014/03/21/e2013sp1-ip-less-dag/

      Exchange 2013 on VMware BPG: http://www.vmware.com/files/pdf/Exchange_2013_on_VMware_Best_Practices_Guide.pdf

      Exchange 2013 on VMware Design and Sizing Guide: http://www.vmware.com/files/pdf/Exchange_2013_on_VMware_Design_and_Sizing_Guide.pdf

      Exchange 2013 on VMware Availability and Recovery Options: http://www.vmware.com/files/pdf/Exchange_2013_on_VMware_Availability_and_Recovery_Options.pdf

       

      Goals:

      Roughly 1100 users, with sizing considerations up to 1250.

      This is a VMware environment and the plans are to deploy Exchange 2013 SP1 on Server 2012 R2 and use the new IPless DAG functionality for redundancy.

      We would keep a lagged copy in our main site as we use our Colo as our primary (sounds backwards I know, but that's just due to the datacenter designs).

      We are going to utilize a mail journal for a single point of reference for any E-discovery or legal hold purposes, we have a company policy where we need to hold email for so long and this seems to be the best method.  We will also utilize an online archive for all employees so they aren't carrying around massive OSTs in the field.

       

       

      Process:

      We used Microsoft and VMware best practices (the VMware BPGs are awesome by the way) and we spoke with a Microsoft Exchange guy who basically verified our solution would be supported (near as I can tell they'll support nearly anything under the sun now).

       

      Our normal server configuration is to retain OS on C: and then any installed software on D: and if it's exchange, database or something else we have all drives for logs/databases/backups etc.

       

      In our deployment we will be using the same mentality, C: for system/OS, D: for the actual Exchange installation, and E: will be the drive letter used for all the volume mount points.

      Any drive with a drive letter is a VMDK.  All mount points located under E: are Windows iSCSI initiator volumes.

      We could have included archive databases on the same volume as well as journal databases, however we decided to break these out for the possibility of different snapshot or replication schedules.

       

      Based on the VMware and Microsoft BPGs we will have 3 multirole VMs.

      As I said before 2 VMs are in our colo as primary DAG members and 1 VM in in HQ as a lagged member for all databases.

      Each VM is 4vCPU and 20Gigs of RAM (this is purely based on the BPG calculations VMware/Microsoft recommend using http://www.spec.org/cpu2006 so make sure to calculate this for your own environment)

       

      Message was edited by: Julian King Sorry just realized I didn't attach the pdf.