Running into more challenges while trying to configure IIB in Active/Active, is this normal?     

I don't think you can configure IIB to run Active/Active with state synchronization betweeen the two IIB nodes. You can setup a HA solution in Active/Passive based on either MQ Multi-instance or with a cluster manager such as PowerHA on the Power platform.

I have good results with several HA solutions in Active/Passive based on MQ Multi-instance, with NTFS v4 or GPFS as the shared file system. Pls ping if you need more info on how this works.

Dixie,

Active/Active would normally mean two Brokers with each one being multi-instance.  So, two brokers and two additional instances.  Typically, each Broker can fail over to the other Brokers server.  Is this what you're trying to do?  What particular problem are you having>  

Regards,

Glen Brumbaugh

What we do is to run multiple brokers on different servers and use a single multi-instance queue manager. Obvoiusly not using associated brokers. The load are balanced between the different brokers using MQ's load balancing and F5 for balancing HTTP(s) traffic. This gives us horizontal scaling.

Amanda,  

It's always challenging talking about HA because there are multiple aspects of availability and the term is often used to describe any, some, or all of those aspects.  In the scenario you described, the Message Brokers are highly available, therefore "HA" for new incoming traffic across either HTTP (due to the Load Balancing).  Any particular HTTP transaction in-flight when a Broker fails will time-out and be the client's problem to handle.  This is typical for HTTP transactions where a time out is always a possibility and so this is usually acceptable in terms of the business requirements for HA.  

It's not clear to me from your description how your Brokers are using MQ.  Are multiple Brokers reading from the same queue?  I would assume so.  In this case, again for new inbound traffic, the loss of a Broker will not stop messages from being processed because other Brokers will be reading from the queue.  this assumes, of course, that the Queue is opened with the "Share" option.  In this design, any message being processed when a Broker fails will be in the middle of an uncommitted Unit of Work.  Therefore, that message will be locked until either committed or rolled-back by the Broker.  Hence it cannot be processed until the Broker that failed is brought back online.  This is the small, but real, value that a Multi-Instance Broker can provide.  Or simply put the broker in a Docker container for "baked in" multi-instance.  

The processing described above may, or may not, meet your business requirements for "HA".  

Regards,

Glen Brumbaugh

Echoing Glen, I think you need to spell out exactly what you want from your IIB Active/Acive configuration. For a start, can you tell us if we are talking about IIB V10 which has considerably relaxed the rules regarding the MQ requirements ? Are you using an RDBMS ? Global Cache ? And do you envisage the two Active instances being geographically close together (supporting HA failover) or distant (supporting DR) ?

And from a business resilience perspective, do you need to provide Continuous Service (rather than Data) Availability ?

As performance specialists always used to say "it depends".