Administering block storage replication

Block storage replication is a form of disaster recovery at the individual volume level, rather than replicating the entire system configuration.

You can configure replication of each volume between local and remote systems in either direction, and in synchronous or asynchronous modes depending on your environment.

Disaster recovery is known as a process of restoring or continuing the use of vital processes after a natural or human-induced disaster, returning your Cloud Pak System environment to a usable state after the disaster occurs.

Typical disaster recovery scenarios might include both the planned and unplanned failover of virtual system patterns, virtual application patterns, deployments, and shared services from a primary system to a disaster recovery backup system. However this method requires you to maintain a secondary backup system and reserve its use only for disaster recovery purposes, and not be able to use the system to support workloads in your environment.

Block storage replication, however, replicates data at the individual volume level, not the entire system environment. There are several advantages to this approach:
  • You do not need to maintain and dedicate a second system in your environment only for disaster recovery. You can use it as another active system, supporting workloads as usual, while also serving as an auxiliary system for replicating volume data from the primary system.
  • You can configure replication between volume pairs in either direction. In this way, both systems can serve as either a primary or auxiliary system for each volume pair.
  • You can define a replication profile to include any number of local and remote volume pairs, each configured to replicate in one direction or the other.
  • Each volume pair can replicate data in either synchronous or asynchronous mode (systems separated by more than 300 km typically must use asynchronous mode to maintain reasonable throughput and response time for deployed virtual machines to which these volumes are attached).
You must complete several tasks to prepare your Cloud Pak System environment for block storage replication:
  • Identify the systems that will be used for replication, in both primary and auxiliary replication roles.
  • Ensure that network communications between the systems are allowed in your network. For more information, see Firewall requirements for block storage replication.
  • Define users on both systems with the appropriate authority to access volumes and perform block storage replication tasks.
  • On each system, create and configure appropriate block volumes or block shared volumes, that are attached to deployed virtual machine instances. volumes eligible for block storage replication must be configured as either Block or Block Shared type.
  • On both systems, deploy patterns as needed to support your workloads, with each system having a unique set of IP addresses.
  • On each system, create a block storage replication profile and define volume pairs for replication. In this process, you specify which volume is primary, which is auxiliary, the direction of replication, and whether replication is performed in synchronous or asynchronous mode.
  • When you define a replication configuration between a volume pair, the associate system name on the source system and its LUN identifier meta data are copied to the target system. This helps identify which target volume is associated with a source volume for disaster recovery scenarios.
    Note: For block storage volume pairs defined prior to Cloud Pak System 2.3.3.x, the associate system name and its LUN identifier meta data are automatically copied to the target system once both the source system and target system have been upgraded to Cloud Pak System 2.3.3.x and the source volume has been assigned an associate system name.
  • Enable and validate block storage replication on both systems, monitoring the replication progress as needed to ensure that primary and auxiliary volumes remain consistent with each other.

To configure your Cloud Pak System environment for block storage replication, two Cloud Pak System systems are required. These two systems can reside in the same data center, but locating them in separate geographic locations offers the strongest safeguard in the event of natural and man-made disasters. Distances up to 8000 km are supported between the two systems.

Each system actively supports workloads, performing the functions and pattern deployment of a typical production Cloud Pak System. At the same time, each system also serves as a block storage replication auxiliary system for the other. Depending on the direction of replication for each volume pair, the system in the auxiliary role receives replicated data from the system in the primary role, and is available for either a planned or unplanned failover. Using this configuration, no data is lost during a planned failover. During an unplanned failover, only data that is in transit from a user of a deployed application, or an unsaved administrative action, might be lost.

For a failover operation from primary to auxiliary systems, you might also configure the replication to reverse the roles between systems after the failover operation completes. The auxiliary system then assumes the role of the primary system, and takes ownership of the volume data, resuming support of the associated workloads and system management data. The original primary system can then assume the role of an auxiliary system, or you might choose to configure a new Cloud Pak System to serve in that role.

The block storage replication process allows the primary system and auxiliary system to have different numbers of compute nodes and different storage sizes. You can replicate from a larger system to a smaller system, with certain limitations.