Multi-site Exchange Redundancy

[Nybbles]

Ok, real quick: Have a client with 2 locations that are geographically seperated by several states. Both sites have a Win2k3 DC and a Win2k3 Exchange 2k3 Enterprise server.

SiteA has a single T1 line and SiteB has a pair of bonded T1's. SiteA and SiteB are connected via VPN by their SonicWall Routers. The lines at either location are also used for Internet access to the workstations at that location.

Basically, the client wants to know what would be required in order to make the mailboxes for both locations avaliable in the event that one of the servers goes down.

Personally, I dont think there's a decent option given the size of the clients business, but perhaps (probably) you guys know something I dont. I think the connectivity between the locations is going to be insufficient for most things.


I was thinking a couple options:

- Set up a couple/few servers in a cluster and run Exchange at a single location (the location with the most local users). Then have the other site access it via RPC-over-HTTP.

- Have the client keep a spare server at each location and make sure to test their backups regularly. In the event of an outage, the new server could be brought up with the OS already installed, install exchange and restore their Exchange data. The nice thing is that the spare server could be a backup in case any other server at either location were to go down as well.

Ideas?

 

[Fint]

How large are the information stores? How much down-time is acceptable? How much lost email is acceptable (i.e., real-time replication, or scheduled replication of information store)

 

[NetJunkie]

Put a spare box at each location and use WanSync by Xosoft. It will keep the backup server in sync with the primary and let you cut over very easily. The cost is about $4K/server.

We use this to keep a backup offsite at our data center. I used to keep the server sync'd over a T1/VPN setup. With 250 active users the backup site was about 4 hours behind on a busy day. But, 4 hours instead of completely down wasn't bad. We've moved to a faster connect to that site and now it's almost always current.

 

[feigned]

- Set up a couple/few servers in a cluster and run Exchange at a single location (the location with the most local users). Then have the other site access it via RPC-over-HTTP.

- Have the client keep a spare server at each location and make sure to test their backups regularly. In the event of an outage, the new server could be brought up with the OS already installed, install exchange and restore their Exchange data. The nice thing is that the spare server could be a backup in case any other server at either location were to go down as well.

Why RPC? You're all ready in a VPN... I guess you could argue the whole less traffic defense.

Suppose you could do a two box back-end cluster at your central point and then front-end a box at the other site. That would be ridiculously easy to setup and even restore at your remote location. What is it, one check-box after Exchange is installed?