Anyone done an Exchange 2007 CCR Geocluster?

oddjob62

oddjob62

oddjob62

Soldato
Joined
8 Nov 2002
Posts
9,128
Location
NW London
I've managed to get the go ahead for this project. Wondering if anyone's done it already?

For those who don't know it's having Exchange 2007 on Server 2008 in a failover cluster setup with the two nodes in different geographic locations (in our case Node A at head office and Node B at our DR site with a 100meg LL in between)
 
nope but do let me know how it goes, it might be something that I think about deploying in the future, my old company did this, but I never took part in the project

Stelly
 
Elric said:
We're putting in an Exchange 2007 Geo cluster, currently running a 2 node 2003 Geo cluster.
Click to expand...

I assume you've currently running with a stretched VLAN then. Are you going to move to the full blown, separate subnets offered by 2008?
 
We are potentially going to be doing this, although it is subject to change. Using Win 2008 to work over different subnets.
 
Well so far i've got it working in my test environment.

2 Physical cluster nodes on different subnets runing the Mailbox Role, and a VM running HT and CA.

Did my first failover today, and apart from the obvious DNS delays, it's all working. Now to try it on my live network and over the 100meg LL to our DR site :S
 
bigredshark said:
Indeed, be interesting to hear how you get on, particularly over 100mb. How are you planning on replicating storage between the two sites?
Click to expand...

No need... Exchange 2007 uses the same log shipping as in SQL 2005 database mirroring to synchronize the t-logs between the live node and the backup node.

2 completely seperate storage systems.

Only thing is the drive letter and folder structure must be identical on both nodes.

The other bonus is that you can run backups on your backup node (even though it's not live), and so the DPM server at our DR site can run continuous backups without any hit on our live server.
 
oddjob62 said:
No need... Exchange 2007 uses the same log shipping as in SQL 2005 database mirroring to synchronize the t-logs between the live node and the backup node.

2 completely seperate storage systems.

Only thing is the drive letter and folder structure must be identical on both nodes.

The other bonus is that you can run backups on your backup node (even though it's not live), and so the DPM server at our DR site can run continuous backups without any hit on our live server.
Click to expand...

Nice and simple setup, 2007 and CCR is a huge improvement on 2003
 
Oddjob, that is almost what we are planning on doing, including potentially putting the DPM server at our DR site (100Mb LES). Uncanny, must be a good way to go then! :)
 
Update:
Ok, so i've started moving into our live environment.
Today weekend i installed the HT/CA server (on ESX Cluster)
Then set up the 2 nodes of the CCR cluster (both at head office at the moment)
Moved a test mailbox over to the cluster (mine).
Tested the failover (works nicely)
Set up DPM to backup the node that eventually will be at our DR site.

This week i will be moving the IT department over to the new cluster, and either this week or next week will be moving one of the nodes over to our DR site.
 
The Mailbox servers are:
Dual Xeon E5410 Quad Core 2.33GHz
16Gig RAM

Roughly 350 users.
Medium usage (not done any proper calcs tbh, we had a budget and so got the best servers we could for that price)

Our current server has only 4gigs of RAM so that alone should give a big boost in performance, especially now that I've fixed the nasty SAN setup that i was given when i arrived.
 
Deathwish said:
Thanks for that. How many disks did you manage to squeeze in or are you placing the mbx on the SAN?
Click to expand...

Got a 12x300gig SAN for each node. Although this is simply because we already have the SAN in place, not because the cluster needs it.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom