If we are creating a data protection strategy, a snapshot is quicker than a split
mirror for the initial create.
Once we have created the split mirror, what will happen is a thing called
incremental re-establishment. We track the changes that are made to the
original logical unit whilst mirror is in the split state.
Now when we re-synchronise the split mirror, we’ll bring it back into the paired
state, its only the changes that get copied over, and then incremental re-
establishment is that much faster. But the initial create can take hours.
Most of the vendors will do this as a background task. They will take a
snapshot and then use the snapshot to create this, tracking changes as they
go. They don’t tie the volume up. It’s still not available (no protection) until that
copy has been done. The best thing to do is create them at the same time.
Split mirrors have to be the same size, and split mirrors can protect you
against primary logical unit failure. But this is still in-system replication.
Snapshots don’t protect you against primary logical unit failure.
Split mirrors do but won’t protect you against subsystem failure.
If your site has failed you’ve still lost access to your data. But because they
are both point in time copies they will protect you against corruption provided
you have one that was taken before the corruption occurred.