Users Guide

7
Asynchronous remote replication
The Asynchronous Remote Replication (aRR) feature allows for a new method of remote replication utilizing point-in-time copies.
This feature supports both Fibre Channel and iSCSI remote array connections. The key features of aRR to consider for SRM are:
• Support for both Fibre Channel and iSCSI remote array replication
• Maximum of four asynchronous remote replication groups (RRG) for each array
• 10-minute sync interval between point-in-time copies
iSCSI remote replication
aRR supports remote replication by using iSCSI protocol, which allows for greater distances for array-based replication at a cost of
latency. Ensure that during datastore creation only data that must be replicated is included on the virtual disks you replicate. Observe
and calculate the value data being replicated and the time required to synchronize the data to determine the expected delay time
between synchronization periods. If the amount of time required to synchronize data is more than the synchronization interval, the
RRG becomes degraded and nonfunctional. Proper sizing of WAN infrastructure is critical for a successful DR solution.
Eects of four asynchronous remote replication
With a maximum of 4 RRGs, all protected Datastore virtual disks must reside in one of the four groups. A group is treated as a single
entity; thereby, when swapping roles, all virtual disks contained in the RRG are changed. If cross replication of Datastores is required
(for example, replicating from recovery site to protected site), the Datastore virtual disks from the recovery site must be contained in
a separate RRG from the virtual disks for the protected site.
Eects of 10-minute sync interval
RRGs require a 10-minute interval between both automated and manual synchronizations. This interval means that requesting a
manual synchronization of the RRG may not occur until after the minimum interval has been reached (10 minutes). This interval may
cause a delay in the SRM workow process, which requires several sync operations to occur for both test failover and failover
workows. The SRA is optimized to avoid requesting a manual synchronization if no changes are detected in the RRG, but if changes
are detected, synchronization is requested. Therefore, you may observe a lack of progress or slow progress through the SRM
workows.
13