MySQL and Message Queue backend for openstack

Solution Verified - Updated -

Environment

  • Red Hat Enterprise Linux OpenStack Platform 4.0

Issue

  • What is the suggestion on mysql and message queue backends?
    • Does Mysql Cluster with Galera works well with RHO 4.0?
    • And do you have a solution on qpid cluster for RHO 4.0?

Resolution

  • Currently Mysql Galera is not supported. High-availablity is available for mysql databases via an active-passive configuration in conjunction with High-Availbility clustering and pacemaker.

  • Red Hat Enterprise Linux OpenStack Platform uses qpid/AMQP as the message bus for intra-service communication. A single message broker would be a single point of failure and a scalability concern. Several approaches have been considered for HA and scaling of the message bus:

    • Active/Passive qpid brokers with a VIP, monitored by Pacemaker/Corosync - This solution provides no message durability or scalability but does provide a reducion in the MTTR for a failed qpid broker. (Note that at present the qpid driver for openstack doesn't enable reliable transfer anyway, so message loss on failure is likely even if replication is used)
    • Federated Brokers - This provides lower MTTR, as this would involve many brokers operating in a grid. A client can chose any broker to send a message through. If the first choice fails, the second can be used. This model provides scalability and HA but not message durability. It also does not work for the OpenStack use case, based on the way that OpenStack utilizes the message bus, unfortunately.
    • Clustered Brokers (Active/Active qpid brokers) - This replicates all messages between all brokers in the grid via Corosync. The negatives of this approach are reduced throughput overall and increased complexity with setup (since Corosync is used) and a scaling limitation of approximately 16 brokers maximum. For the time being, this scaling limitation is not a known issue.

This solution is part of Red Hat’s fast-track publication program, providing a huge library of solutions that Red Hat engineers have created while supporting our customers. To give you the knowledge you need the instant it becomes available, these articles may be presented in a raw and unedited form.

Comments