I think the real question is, how much can you afford to spend?
The entry level DR features of SQL Server like log-shipping work ok, then database-mirroring is worth looking at.
If your budget will stretch far enough, a geographically-dispersed cluster running on a mirrored SAN is pretty much the Rolls-Royce of DR \ Contingency architecture.
My experience is that with decent kit, the failover time is in the region of a few seconds, and if you have your servers in different geographic locations, you can recover service from complete site loss before anyone has actually noticed.
That said, for my money the best and cheapest DR solution is always a well tested backup regime.