Image Support #3307
MQ is not processing messages fast enough as expected
|Image:||WebSphere MQ 8.0||Your Marketplace Account ID:||0000-0000-0000|
|Operating System:||Linux||Marketplace:||Microsoft Azure|
|JRE:||v1.8 JRE/JDK 64-Bit||Customer State:||Elkhorn|
|Instance Type:||Not Applicable||Customer Country:||US|
We need your help in Raising PMR with IBM with severity-3 as we suspect that messaging queues are not processing messages fast enough at MQ server level.
Below is our architecture.
We are using MQ v8.0 installed on one separate VM (linux machine).
Our application is installed on 6 different VMs and MQ-client is also installed on these 6 VMs.
Application is connecting to MQ server using these MQ-client.
Below is our test case:
Test 1: Application VM 1,2,3 -> SPDH001 queue -> NF31.IS queue -> HISO31 queue -> asset 1
Test 2: Application VM 4,5,6 -> SPDH002 queue -> NF51.IS queue -> HISO51 queue -> asset 2
Now, when we are running any one of above test, we are getting 4k transaction per second (TPS). However, when running both we are getting only around 5K whereas ideally it should be around 8K. Everything looks good on the SPDH side. We are suspecting some bottleneck is there in the issuer side between application –> MQ –>Asset and back. When running the test with all VMs, we are observing the Grinder server is processing around 3 MB/s. Now for each transaction, we have 8 MPUTs and MGETs so that would be 24 MB/s in the MQ from the SPDH side. In the Asset side, the message is much longer and it should be around 6 times the SPDH message size limit. So this might be a tentative bottleneck in terms of bandwidth usage of the MQ.
- Status changed from In Progress to Marked for Closure
I have been informed you already got some solution for this problem.
Please let me know if there's anything else we can do for you, otherwise we'll timeout this ticket in a few days time.
Remember you can always open a new ticket if you need any further assistance!