We are evaluating approaches to DR. For SQL Server we have the usual log shipping, replication, and various 'AlwaysOn' methods. The other approach is to use one of the Hypervisor or SAN based disk level replication technologies.
Certainly, one advantage of the SQL approaches is the the system at the remote DR site is 'warm' or even 'hot'.
But I have not been able to find discussions on the relative efficiency of the two approaches. I would imagine that any Hypervisor or SAN based replication needs to replicate ALL the disk writes. That includes the MDF as well as the LDF writes etc., AND any Windows NTFS writes such as the system journal and file attribute structures if the file grows or shrinks. Whereas the SQL methods are really only replicating the resulting DB change, and letting the remote system apply the change.
I would think that the SQL replication could therefore use significantly less network bandwidth between the primary and DR sites.
Does anyone have any experience or opinions here???