Ensuring Data Consistency in Asynchronous Mode Replication

Starting in the Sun Cluster Geographic Edition 3.2 11/09 release, Hitachi Universal Replicator can provide guaranteed data consistency in asynchronous mode replication, in which the replication fence level is set to async. Asynchronous mode replication is commonly used between a primary data center and a distant disaster recovery site. Guaranteed data consistency in asynchronous mode is therefore critical to the functioning of a disaster recovery system.

Guaranteed data consistency in asynchronous replication mode requires the following:

• You must run Hitachi Universal Replicator. Hitachi TrueCopy cannot always guarantee data consistency in asynchronous mode.

• On both clusters of the Sun Cluster Geographic Edition partnership, you must have Hitachi storage arrays that are supported for use with Hitachi Universal Replicator. Talk to your Sun representative for a list of currently supported hardware.

• You must configure journal volumes on the Hitachi storage arrays at both sites. For instructions, see the Hitachi documentation for your array.

• A journal volume must be associated with each asynchronously replicated paired device in the /etc/horcm.conf file. You configure this association in /etc/horcm.conf as a property of the parameter HORCM_LDEV. You cannot use the property HORCM_DEV. For details, see Configuration of the /etc/horcm.conf File and Journal Volumes.

• Each asynchronously replicated Hitachi device group that is used by one particular service or application must be assigned the same consistency group ID (CTGID) as the protection group that manages it. To do so, you can complete the following steps:

  1. Create the protection group with the CTGID that you want to use.

  2. Add uninitialized Hitachi device groups to the protection group.

  3. Start the protection group.

  For details, see Ensuring Data Consistency for Hitachi Universal Replicator in Asynchronous Mode.