If a link is configured to use asynchronous snap-based replication for it's initialization, can that link later be changed so that snap-based replication is not used?
In this specific situation I have a large, 15Tb lun replicating to a site >200mi away, with only a 50mb connection. We have seen a lot of packet loss and high load events and are trying to get a full initialization to complete but it never seems to complete. Other CG's are replicating just fine.
Looking at using snap-based replication as a way to "pre-seed" the data and then switch back to normal asynchronous replication.
I would recommend upgrading to 4.1.2.1 (4.1 SP2 P1) which went GA recently.
Snap-based replication will not help if WAN is the bottleneck since the initialization process is that same from a WAN perspective. if the production storage is slowing initialization so IO throttling can be configured to limit the read bandwidth RP consumes during initialization (config_io_throttling command in admin CLI) or indeed snap-based replication which can help since RP replicates from a static image w/o having to replicate wrote to production volumes during initialization.
Snap-based replication initialization is similar to the one in normal Async. So, both will replicate over the WAN, assuming that's the problematic component.
Idan
4 Apprentice
•
675 Posts
750
0
Posted August 3rd, 2015 05:00
Jeremy,
I would recommend upgrading to 4.1.2.1 (4.1 SP2 P1) which went GA recently.
Snap-based replication will not help if WAN is the bottleneck since the initialization process is that same from a WAN perspective. if the production storage is slowing initialization so IO throttling can be configured to limit the read bandwidth RP consumes during initialization (config_io_throttling command in admin CLI) or indeed snap-based replication which can help since RP replicates from a static image w/o having to replicate wrote to production volumes during initialization.
Regards,
Idan Kentor
RecoverPoint Corporate Systems Engineering