Topic
3 replies Latest Post - ‏2012-10-14T03:39:38Z by fjb_saper
NickLaqua
NickLaqua
6 Posts
ACCEPTED ANSWER

Pinned topic WMB versus standalone WMQ

‏2012-10-08T02:47:02Z |
Hi there,

we are reviewing our messaging and integration topology and considering whether to discontinue our legacy standalone MQ hub in favor of rolling it into the WMB platform, given that WMB comes with WMQ anyway. Main reasons are support and provision simplicity as well as cost savings.

Is there high level best practices when to maintain a standalone MQ versus using a centralised "ESB" platform for all (realtime) integration needs ?

thx Nick
Updated on 2012-10-14T03:39:38Z at 2012-10-14T03:39:38Z by fjb_saper
  • fjb_saper
    fjb_saper
    120 Posts
    ACCEPTED ANSWER

    Re: WMB versus standalone WMQ

    ‏2012-10-09T07:35:08Z  in response to NickLaqua
    Well, it really depends and you may or may not see any savings from the new topology.

    Typically the "legacy" apps use MQ as a ramp to the ESB, and it will make sense to keep some MQ licenses around for them to use.

    Looking at the topology I would recommend that applications not attach directly to the broker queue manager. This will allow you to build an MQ Cluster with WMB and use no single point of failure. You may also factor in the use of HA.

    Anyways your statement is way too broad to be answered with any accuracy. And in any circumstances, your mileage will vary...
    • NickLaqua
      NickLaqua
      6 Posts
      ACCEPTED ANSWER

      Re: WMB versus standalone WMQ

      ‏2012-10-09T10:01:40Z  in response to fjb_saper
      Thx for the response.

      I agree that my question was rather open-ended and I didn't really expect a "black and white" conclusive statement but rather some advice, aspects and general pro's and con's that should be considered before making the decision.

      The key advantage that I can see with a single combined platform is reduction of complexity for mediated message transfers as you don't have to maintain a MQ network integrating the standalone cluster with the broker queue manager. Generally, all things being equal, less moving parts are always preferable.

      Disadvantages, on the other hand, are the tight coupling between messaging (WMQ) and integration (WMB) as they have to be co-located. So there might be an impact in terms of independent scalability. You mentioned HA, my assumption was that a WMB cluster (incl. MQ) can be highly available too.

      One perspective could be to treat WMQ as a MQ protocol "adapter" for WMB, in the same way that WMB has an inbuilt web server which acts as the HTTP protocol "adapter".

      So maybe you can point out a couple of non functional aspects and/or technical constraints why connecting apps directly to the broker QM is not a good idea.

      Many thanks,

      Nick
      • fjb_saper
        fjb_saper
        120 Posts
        ACCEPTED ANSWER

        Re: WMB versus standalone WMQ

        ‏2012-10-14T03:39:38Z  in response to NickLaqua
        Let's say you do load balancing via an MQ Cluster (remember MQ cluster is not HA).
        Direct connection to the broker qmgr would bypass the load balancing.
        It would also prevent you from removing any single broker from the cluster for instance to perform maintenance and such...

        Think what will provide you with the most independent scaling, and be completely end user transparent...