TechOnTip Weblog

Run book for Technocrats

Best Practices for Exchange 2007 CCR Stretched Cluster

Posted by Brajesh Panda on March 8, 2010

In course of my SI Projects I found these are should be the best practices for a Exchange CCR Stretched Cluster. Let me know if anything I should include in my list.

Decision Making Factors

  • Is automated failover is required in unplanned site outage?
  • How frequently you are facing situation like complete site disaster?
  • Where is the major user base located? Is at Prod Site or DR Site?
  • Do we have enough bandwidth between these two sites?

     

AD Sites

  • Both CCR Hosts must be in same AD Site

    N.B:    Typically, Active Directory communication is controlled based on the AD Site design. Since both datacenters are using the same AD site, there is a risk that servers and clients in 1 datacenter may be communicating to domain controllers in the secondary datacenter. Cross authentication may leads slow response in Directory Server Access!

    –    Plan two domain controllers in each site

Network

  • Encase of Windows 2003 Server we have stretch the subnets across datacenters
  • Windows 2008 Stretch cluster can span across different subnets; however we have to consider below points;
    • Latency must not be more than 50ms
    • Two different subnets should be available for Cluster Private & Public Communication

 

Enough bandwidth

  • If you have large number of transaction oriented mailboxes you may need big bandwidth pipe between two sites.
  • Temporary enough back is required at the time of backup/restore operation
  • Dedicated Dark Fiber or Gigabit network is suggested between two buildings.

 

Quorum Location

  • CCR cluster use File Share Witness Quorum Model for operation. For a stretched Cluster a 3rd site is recommended to Hold the Quorum Folder to automated complete site disaster
  • Otherwise manual forceful commands will be used to bring up the cluster which is not recommended

 

Domain Name System

  • AD Integrated DNS must be used because failover will change the address of clustered mailbox server. So new subnet IP must get registered dynamically
  • Default DNS TTL for host records is 20mins & minimum time for AD replication is 15min. Hence to replicate your changed records quickly you have to reduce the TTL value. Hence DNS Host record will expiry quickly creating replication.
  • GPO can be enforced to clear DNS Cache from Windows Clients quickly but it may impact your DNS infrastructure. Better keep this value little more than TTL value.
  • Even you can configure GPO to configure your Outlook time out time. Better keep this value little more than TTL value.

 

Transport Dumpster

  • It provides facility to recover mails on transport queue which are not yet committed to 2nd Node in case of 1st Node failure
  • If we set Transport Dumpster in the 1st site in case of site unplanned disaster none of those mails will be recovered. Have to take this risk 😉

 

Exchange Features

  • Since the mailbox role needs to have a HUB and CAS server available in the local AD site, this means that we need to place additional hardware in both datacenters to support the HUB and CAS role. This is in case 1 datacenter fails – the other can continue with all roles.

 

Mail Flow

  • Mail Submission Service automatically do load balance between the Hub Transport servers that are located in the same AD site as the Mailbox Server.
  • In this scenario the Mailbox Server could be talking to a HUB server in the remote datacenter. This could generate additional network traffic and make it more difficult to control scheduled outages (i.e. HUB server reboot could impact production mailbox servers).

 

Client Access

  • Exchange 2007’s Auto discovery service will automatically detect which Client Access server is closest to the user’s mailbox based on AD site design.
  • If it selects the CAS server in the secondary datacenter, this would create additional network traffic and might generate client performance issues
  • Use hardware load balancing solution to load balance all CAS servers because they are located in different subnet & Windows NLB doesn’t support that. Else you can use DNS Manual CNAME change method.
Advertisements

One Response to “Best Practices for Exchange 2007 CCR Stretched Cluster”

  1. As a matter of fact, the controversy in a way added to their overall popularity.

    Another fascinating stop for vacationers is Pegae Hill, which is
    a popular camp website of the North Korean People’s army during
    their fight against the Japanese. facility in Kingsland, GA, where we have completely rebuilt
    all of the soundstage of the original “Star Trek.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: