For testing purposes we used SRDF to duplicate the DB and logs and
successfully hot split the R1 and R2 several times. The DB always went
into recovery mode, but would start. If you split all of the R2's at
the same time it is the same as crashing the host, not the best thing,
but workable.
If yo
I have a customer that has done this successfully with SVC (IBM product)
replication.
The trick I believe is to have the DB and logs in a "consistency group",
which means that SRDF must enforce the correct order of writes.
Here is the TSm doc on what is required:
http://www-1.ibm.com/support/docv
