Recovery Exchange 2000 Servers

[dvientos]

This is the situation. Currently have one exchange 2000 server with one organization and one store. The public and private database is a combined 10GB and is growing but not exponentially. We want to create a 2nd exchange server that can take over as the exchange server in the event that the oringinal server dies. In essence a Recovery Server but we want to be able to restore the database to the 2nd server and then go ahead and have users connect.

Along with that we have a SAN that we can use. I want to be able to use the SAN and move the databases to the SAN partitions that are available and then be able to have the second server either use the same database or at least be able to restore from the database.

I have explored the cluster option and that is not a viable option due to the astronomic costs.

Another option I am exploring is creating a new forest and building a new exchange server with the same config and nameing of the server/organization/group etc. But the issue I have with this is that I do not have access to the users that are on my other forest with the live server. I would have to create each user on the new AD and maintain it constantly. And even if I could do this I can't move the server object between forests. I thought of maybe building the server to the point where it has a blank database and then leaving it alone. Then in the event of any emergency I can physically take that server which is an exact duplicate of the orginal exchange server and change the domain or rather have it join the domain where the orginal server was. BUT then there will be issues with the SID's not matching on the AD. So I am at my wits end.

Am I forced to go with a cluster? We have a 60 min window where we can be down after that I am pretty much out of job once the servers are recovered.

Any suggestions are welcome.

 

[wallst32]

Install the 2nd server with Windows and the same service pack and patches, but not Exchange. If you ever need to use it, rename it to the same server name as the original, join the domain, restore your Exchange backups to it, and then run Exchange setup in diaster recovery mode and add the service pack. I'm assuming your AD is installed on another system that is fine.

Other options will have a higher cost. Also, there are different ways to combat different points of failure. A cluster protects you against hardware failures, and a bad OS install, but not against corrupt IS. Point in time snapshots of the IS would be the quickest way to restore a bad IS. There are also 3rd party packages that allow you to mirror the IS between two servers in realtime and allow you to switch back and forth between them, but they are very expensive.

If you have 60 minutes, you need to explain to your boss whether or not that is achieveable with the resources you've been given (for example, I have older hardware - a Windows and Exchange install alone exceeds 60 minutes). I would think you need to do point in time snapshots and hardware redundancy to meet that goal. Also, you need to keep the IS size down, either by enforcing quotas, or splitting mailboxes over multiple stores.

 

[dvientos]

Hmmm Actually now I am leaning towards a product that I saw on here called neverfail. Seems to do what I want as far as hardware/software failover. With the cost of Microsoft's clustering solution (close to $50k)this solution would just require me to invest $9k with hardware instead and I get all the things I am looking for.

I need to do some more research on the company and the product.

When it comes down to it the head honcho wants to have the nice warm fuzzy feeling of a cluster failover at the cost of just buying another server and restoring to it. Bottomline is the server dies it will take longer to install and configure not to mention how things never work right the first time anyway and that sickening feeling all admins get when the restore starts and you are just waiting for something to go wrong because the backup was bad.