MQ

MQ

Join this online group to communicate across IBM product users and experts by sharing advice and best practices with peers and staying up to date regarding product enhancements.

 View Only

  • 1.  MQ in Openshift

    Posted Fri October 25, 2024 05:38 PM

    Hi!

    Anyone here  had success migrating from the IBM MQ virtual machine model to the container model, for example OpenShift?



    ------------------------------
    Renan Vasconcelos
    ------------------------------


  • 2.  RE: MQ in Openshift

    Posted Mon October 28, 2024 09:30 AM
    Edited by Jim Creasman Mon October 28, 2024 09:33 AM

    We didn't migrate from another MQ installation, but started from the beginning with a cloud deployment of MQ.  We have it running in our private cloud (not OpenShift).  If you're running in OpenShift then IBM provides operators that can be used to manage the deployment.  Otherwise, they provide sample Helm charts you can customize to work with your cloud provider.  That is the path we took.

    We have multiple MQ uniform clusters deployed into our cloud.  These have been providing messaging services to our production services for several years now.  Deployment to a cloud does require persistent cloud storage for the MQ server file system (unless you don't care about message persistency ;-)).  The MQ servers run as stateful sets and we also host a CCDT service that provides connection information to the clients for each cluster.  This is how our typical configuration looks.  Are there specific questions you have?



    ------------------------------
    Jim Creasman
    Enterprise Modernization Architect
    Primerica
    Raleigh NC
    ------------------------------

    Original Message


  • 3.  RE: MQ in Openshift

    Posted Fri November 01, 2024 03:37 AM

    Yes, we did migrate our MQ platform from AIX in our private data centers to OpenShift (in Azure) in an MQ Native HA setup, using the IBM Operators. We also did migrate IIBv10 platform to ACEv12, also in OpenShift.

    We are in production now for almost two months. So you may call it a success, however.....

    We did struggle a lot to give our customers the same or even a better performance experience than we had with AIX.
    But because AIX is a very stable platform, there were never unplanned failovers, so the customers never experienced downtime, accept in maintenance windows.

    In OpenShift however, they do experience downtime, when a queue manager goes down (more than 30 seconds). 

    In the next months we will try to fix it to address the near zero downtime requirement, which Native HA does not offer.
    E.g by combining it with uniform clusters? If anyone has experience in combining Native HA with clustering, we would like to hear from them. Maybe share a diagram with their setup? 



    ------------------------------
    Gerlof Bril
    Integration Infra Architect
    Achmea IT
    Netherlands
    ------------------------------

    Original Message


  • 4.  RE: MQ in Openshift

    Posted Fri November 01, 2024 01:04 PM

    30 seconds is a VERY long time to fail over for any of MQ's HA environments.

    How much of the 30 seconds downtime is the time taken to recognize that an issue exists, and how much is the time taken to actually fail over, reestablish the connections, and resume the workload ?

    Is a planned failover vastly faster than 30 seconds ?



    ------------------------------
    Andrew Hickson
    ------------------------------

    Original Message


  • 5.  RE: MQ in Openshift

    Posted 11 hours ago

    I migrated the queue manager from one node to other so according to my rough estimation we can place the queue manager config file in the pod and we can restore the queues to its state.



    ------------------------------
    Arslan Ali Naqvi
    ------------------------------

    Original Message


Related Content