Active-Active replication provides a zero-RPO and zero-RTO solution however it is not a DR solution. Async has a lower latency for applications, since there is no in-band data flow to the target. Active-Active replication provides a higher resilience, but requires a very low latency between systems. As a result, the systems cannot be far enough from the primary site to provide a DR solution. 

 Concurrent replication adds asynchronous (async) replication to a DR site; if the primary site fails, there will always be a DR copy available.

This solution is based on the existing async and Active-Active replicas and leverages the pre-existing advantages in the InfiniBox Replication portfolio. 

The Async Concurrent Replica, has benefits similar to a regular async replica: 

  • Supports RPO as low as 5 seconds, providing a low near-sync replication rate.

  • A very fast and effective non-disruptive snapshot (based on InfiniSnap).  

  • Optimized for TCP/IP, this protocol is designed to function in both stable interconnects, and noisy/unstable ones.

  • Since most replication operations are performed in the RAM, there is a low performance impact. 

  • The DR copy can be tested at any time by taking a writable snapshot on the target

The systems at the primary site are responsible for the async; replication, as shown in the following diagram:

Managing the replica

Create a replication link between one or both of the Active-Active system and the DR system to manage the async replica. 

For more information regarding the network setup, see InfiniBox Best Practices Guide for Setting Up the Replication Service.

Supported datasets 

Version 5.5 only supports Consistency Groups (CG).

If a single volume needs a 3-site replica, it must be placed in a CG.

Creating the async concurrent replica

Before creating the async concurrent replica, you must make sure that the Active-Active replica exists on the replicated dataset.

Once the dataset is replicated according to the Active-Active standards, the only additional requirement is create create an async replica on the same dataset.

Creating a replica in InfiniBox can be done using the GUI or the CLI: users need to select the following:

  • Replication type
  • Local dataset (source)
  • Remote system
  • Remote dataset or pool.

InfiniBox replicas can be created in two ways, depending on whether the dataset is new or existing:

  1. Creating a new replica
    Select a pool on the remote system.
    A remote dataset is automatically created on the remote system in the selected pool, and the replica is created between the source dataset and the new remote dataset.
  2. Select an existing replica 
    The user select a target dataset on the remote system.
    The replica is created between the source dataset and a remote dataset.

Once the replica is created, the source dataset will start replicating the target dataset.

Deleting the replicas

To delete the replicas:

  • First delete the async concurrent replica.
  • Next delete the Active-Active replica on the same dataset.

When a replica is deleted, it is possible to maintain the the replica staging area. The staging area is a snapshot containing the last consistent data that was replicated between the datasets, and can be used when re-creating the replica later on to avoid the full initial sync.

Infinidat highly recommends keeping the staging area in the event that these snapshots need to be used for recreating a replica. 

Async replica operations on a concurrent replica

Concurrent replica supports the following async replica operations:

  • Sync now
  • Suspend replica
  • Resume replica
  • Update RPO and interval

The following async operations are blocked on concurrent replicas:

  • Change role
  • Resizing the datasets
  • Add/remove member from a CG

Failure Handling

DR link failure

A DR link failure will affect the async replica only.

If the link fails, the async replica will pause the sync job that is currently running. Once the link goes back up, the sync job will automatically resume and the replica will continue to transfer data according to the settings.

Active-Active failure 

In case of an active-active failure there are 2 behaviors that will occur based on the concurrent replication setup:

  • If the concurrent replica is defined on the Active-Active replica's available dataset, the concurrent replica continues replicating even if the link is down.
  • If the concurrent replica is defined on the Active-Active replica's lagging (offline) dataset, the concurrent replica will complete the current sync job, but no new sync jobs will be created since the dataset has no new data from this point.

The volumes remain active on one of the systems and will lag. If the async replica is defined on the active side, the async replication will continue. If the replica is defined on the side that is not active, the replication will pause.

Infinidat recommends creating the async replica on the preferred system.

The Active-Active replica will manage the failure as defined in Active-Active replication, regardless of the concurrent replication defined on the dataset.

Was this article helpful?
0 out of 0 found this helpful

0 out of 0 found this helpful

Last edited: 2022-08-06 08:14:06 UTC