Policy-Based Replication

Policy-Based Replication is a method of storage replication available on IBM Spectrum Virtualized-based Storage Systems such as IBM FlashSystems. Policy-Based Replication provides simplified configuration and management for storage replication by applying a policy to a set of volumes in a Volume Group. The policy applied to the Volume Group defines the attributes for the replication, such as the type of replication or the recovery point objective.

Asynchronous Policy-Based Replication

Asynchronous Policy-Based Replication provides asynchronous replication between two Storage Systems, for all the volumes in a volume group. When you apply an asynchronous policy to a Volume Group, the storage system provides the following advantages automatically:

  • Provisions the Volume Group and matching volumes on the remote storage system.
  • Establishes asynchronous replication between Volume Group volumes and their matching counterparts in the remote group.
    Note: Asynchronous replication resembles the Global Mirror with Change Volumes solution. However, the storage system manages the change volumes and coordinates as necessary to make sure that the Recovery Point Objective remains as low as possible, under the value set in the policy.
  • Provides notification when the Recovery Point Objective is higher than the wanted value set in the policy.
  • Provides easier management in failover and failback scenarios.
Figure 1. Async Policy Based Replication
Note: The IBM Copy Services Manager server detects any Volume Group with an applied Asynchronous Policy across all connected IBM FlashSystems or IBM Spectrum Virtualize-based systems. Upon detecting a Volume Group with an Asynchronous Policy, the server creates an Async Policy-Based Replication session and populates it with the volumes that are contained in that volume group. Once the session is created, the IBM Copy Services Manager server can then be used to monitor and manage the replication, either through manual commands or Scheduled Tasks.
This graphic illustrates how Async Policy-Based Replication functions for FlashSystems, IBM Spectrum Virtualize, or SAN Volume Controller Storage Systems.
For Flash Systems, IBM Spectrum Virtualized, and SAN Volume Controller Storage Systems, each copy set in the Async Policy-Based Replication session consists of two host volumes.
Note: On newer FlashSystems, Asynchronous Policy Based Replication is now supported as part of the support for Storage Partitions. Multiple partitions can be defined and used to manage the Asynchronous Replication. The Asynchronous Policy Based Replication session will automatically detect when the volume group is managed in the partition and display the upper and lower partitions for the volume group.

Asynchronous Policy-Based Replication with High Availability

Starting with the FlashSystem 8.7.1 release, you can now setup a Storage Partition for High Availability. Volume groups with Asynchronous Policy Based Replication policies can then be made Highly Available as well as Disaster Recovery capable. When the Copy Services Manager Asynchronous Policy Based Replication session detects that the partition and volume group have been setup for High Availability, the session will indicate this on both the Session overview as well as the Session Details panel.

On the Session Overview, the HyperSwap column will indicate "Yes" if the AsyncPBR session is highly available. If the session is not highly available, the column will be blank.

On the Session Details panel, the session has a field called "H1 HA Status". This field will indicate the current state of high availability. It will indicate "Enabled" when High Availability is setup and the volume group is protected.

If an issue is detected with High Availability on a session, the H1 HA Status will indicate "Problem". When the H1 HA status is problem a status message will appear on the session and the session will be in a Severe status.
Note: When an Asynchronous Policy Based Replication session is running in the H2 to H1 direction, no High Availability is supported. The H1 HA Status will indicate "Problem" but the session will remain in a Normal status as this would be the expected state for high availability.