On 8/25/2011 7:46 AM, Jon Hill wrote:
Is it possible for a snapmirror update command (one not mediated by SnapDrive) to temporarily halt I/O on the source disk? More to the point, is it possible for a snapmirror update command to harm the source server if that server is an active Exchange 2003 box?
We do own SnapDrive for Exchange, but installing it is not an option at this time.
This snapmirror solution is being suggested as an emergency solution to replicate our Exchange database for the next few days, while our operations team tries to figure out why DoubleTake has failed for four days in a row. They understand that without SnapDrive’s VSS goodness there is a good chance of data loss on the target end, but they’d like to give this a shot on the theory that it’s better than nothing.
But it’s only better than nothing if it doesn’t interfere with the production (source) server, hence my inquiry. If there is any chance of an I/O suspension that could take the production server offline, then obviously this is a nonstarter.
snapmirror initialize, update, etc.. will not halt io on the source, it will however impact network (if they run over the same path, but this can be minimized) your snapmirrored copy will not be in a consistent state (am assuming you are using luns) on the dest side you will then have to do some steps to restore the exchange.