We are currently replicating using SRDF/A, due to some cache and disk utilization issues on the target side, we are encountering SRDF/A drops, so we are planning to implement Recoverpoint replacing SRDF replication.
My question is , whether a full sweep of volumes is required after implementing CRR? or a short initialization would suffice , picking up from where the SRDF has left?
An initial full sweep is required from the RP. This is because the the metadata information about track movement between source and target is completely different in SRDF versus RP, and is not shared between the two within the Symmetrix. The RP has no idea what data was transferred to the target before the implemtation occurred.
Thanks Chris. Good to know but bad for us we are constrained on the link BW. Probably have to bump up the link temporarily to get this fixed.
RP has throttling capabilities during full sweep. Have a read on page 339 of the Administrator's Guide. Including the link below:
But I don't think throttling IO can help here, setting throttling to high will further increase the initialization time?
Yes, you will increase the time to init. However, if you have BW limitations, there's going to have to be some sort of trade off. If the constraints might lead to saturation, and you need to avoid saturation, the throttling will do that trick, but naturally, the init will take longer.