Disaster Recovery Support in the IBM(R) MQ Appliance 22.214.171.124
John_Colgrave 120000E4PG Comments (2) Visits (5743)
A new level of firmware, 126.96.36.199, has just been released for the IBM MQ Appliance. You can find it here: http
In addition to fixes for a number of issues, this level of firmware includes the first phase of the support for Disaster Recovery (DR) in the IBM MQ Appliance.
At least two IBM MQ Appliances are required to use the DR support. It is expected that one IBM MQ Appliance will be deployed in a site where applications normally run. We refer to this as the main site. The other IBM MQ Appliance will be deployed in a site that will take over from the main site if a disaster occurs. We refer to this other site as the recovery site.
When we talk about the current role of a queue manager configured for Disaster Recovery we talk about the Primary Role and the Secondary Role. A queue manager can only be started when it is in the Primary Role. I will simplify this to talk about the Primary instance and the Secondary instance of a queue manager.
The initial state for a typical scenario is that the Primary queue manager instance will be on an IBM MQ Appliance in the main site, and the Secondary queue manager instance will be on an IBM MQ Appliance in the recovery site.
Why HA and DR?
The initial release of the IBM MQ Appliance had support for High Availability (HA) and this support was improved in 188.8.131.52 and 184.108.40.206. The HA support is designed to provide very fast automatic failover with no loss of data, and to be deployed within a single datacentre, ideally with the two appliances necessary for HA connected directly to each other.
We always intended to provide a different solution for DR as the requirements for DR are different to those for HA, including:
With 220.127.116.11 you now have a choice of using HA or DR depending on your requirements.
Differences between HA and DR
The main differences between the support for HA and DR in the IBM MQ Appliance are:
The simplest DR scenario is that a DR queue manager is configured and when the main appliance fails (and will not be restored sufficiently quickly) the queue manger is started on the recovery appliance, possibly as part of a broader Disaster Recovery process.
The commands to achieve this are as follows.
This scenario demonstrates the use of some new commands specific to DR and some existing commands that have been updated to support DR.
You can also use the IBM MQ Console to configure DR queue managers.
The details can be found in the Knowledge Center for the IBM MQ Appliance: http
Replication, synchronization and snapshots
Data is usually sent from the Primary instance of a queue manager to the Secondary instance of the queue manager. There are two modes in which the data can be sent:
If there were to be a problem with the Primary instance or the network while synchronization is in progress, it would not be possible to start the queue manager, as we could not guarantee that we had consistent data. To avoid this problem we take a Snapshot of the consistent state of the queue manager data at the time that the synchronization process starts. If there is a subsequent problem, it is possible to revert to this known consistent state and start the queue manager on the recovery appliance, but without any of the updates that happened on the Primary instance while the network was disconnected. To allow for a snapshot to be created for each and every DR queue manager, it is necessary to keep enough free space for a complete copy of the data for every DR queue manager.
Limitations in 18.104.22.168
There are some limitations with the DR support in 22.214.171.124.
The main limitation is that an individual IBM MQ Appliance can support HA queue managers or DR queue managers but not both.
Another limitation is that we have tested the DR support with simulated network latencies of up to 100ms round trip times so at the moment support is limited to environments where the round trip time is no more than that.
Further details on the limitations for DR in 126.96.36.199 can be found in the Knowledge Center: http
Find out more on GitHub
There is a GitHub repository dedicated to the IBM MQ Appliance as part of the IBM Messaging organization on GitHub: http
I am going to start adding more information regarding the DR support there so keep checking for new content.
Using the MQ Console
Update: This blog post shows how to configure DR from the MQ Console.