need a fault tolerance exchange server 1

[bllk]

I have been given the task of providing fault tolerance for our single exchange 5.5 server. I was wondering if anyone could give me any hints at what would be reliable and relatively inexpensive. Also is it possible to mirror mailboxes on two different exchange server without installing clustering?

 

[SimonDavis]

On the server itself, you have the same choices as a normal NT server - raid options will behave on the exchange server in the same way as any other server.

You can install a second server and have the two replicate each other, although this is definitely not the cheapest option.

It kind of depends what kind of fault tolerance you want - do you want to avoid downtime, or protect the data? (of course using fault tolerance in place of a good backup will always end in tears - but now I sound like your granny).

 

[cpurom]

What kind of hardware is Exchange running on now? Are you looking for fault tolerance or load balancing? Is the current server "pegged" to capacity right now? What's your definition of "inexpensive"? (There is no such thing as inexpensive fault tolerance, aside from a good nightly tape backup.)

What's the absolute longest time you can suffer the server being down?

Microsoft recommends the following configuration for an exchange box with fault tolerance:

1 set of mirrored drives for the operating system partition
1 set of mirrored drives for the MTA
and
1 set of Raid 5 drives for the message database.

That's (7) ultra-scsi drives, at a cost of about~~$400-$600 per drive(?)
plus a good hardware-based raid controller~~$800 (?)

 

[Brent71]

It depends on what sort of fault tolerance you want. MCS is about the only product out there that will provide fault tolerance for the actual application(Like Exchange). However, if you want a means to protect your Exchange server without installing MCS, then there are a few replication software packages out there. I'd try and look for something that had easy instructions to set up the exchange server and didn't require rebuilding the server to make it fault tolerant. My experience has been that most servers require a script or batch file to help make the switch to the fault tolerant server. Look at the script. My experience has been, that the longer a script is the more likely it is that it could have errors. Another thing to think about is if the fault tolerant application is going to put a load on your production server. Many of the fault tolerant applications require that the secondary server have the data committed to disk before the primary will continue writing. This can obviously slow things down. NSI Software, Legato, and Veritas all have multiple products that do what you want.