Hi
before jumping to try a fix, I would recommend trying to understand the issue first.
From your description, both configurations (fast and slow performance) are using MQ Client.
MQ Client is sensitive to both network bandwidth and network latency.
I believe it would be sensible to understand the difference in network latency and bandwidth between the "same cluster" (AKS-AKS) configuration and the "distant cluster" (AKS-ARO) configuration.
If the latency is significantly different, you can then discuss options with your provider. If the latency is similar then at least you know something that you don't have to worry about and you can start looking elsewhere. If your messages are large then bandwidth could also be an issue, but I would expect bandwidth to be fine for most types of applications. Worth checking it though if your messages are quite large.
But at the moment, I don't think you have enough information to make sensible decisions about what to try to do to fix your issue.
In theory, ping might be able to indicate the round trip latency between endpoints, but in practice the icmp echo packets generally get dropped in cloud based networks. You may be able to use a tcpping tool instead.
Bandwidth can generally be measured with reasonable accuracy with a file transfer tool.
Regards,
------------------------------
Neil Casey
Senior Consultant
Syntegrity Solutions
Melbourne, Victoria
IBM Champion (Cloud) 2019-22
------------------------------
Original Message:
Sent: Wed October 18, 2023 03:40 AM
From: Brajendra Kumar
Subject: performance issues
Thanks Colin Paice . ACE is on Azure Kubernetes Service (AKS) while MQ is installed Azure Redhat Openshift platform .MQ and ACE are different system/box.
could you please advise us what network configuration need to change.
------------------------------
Brajendra Kumar
Original Message:
Sent: Wed October 18, 2023 01:51 AM
From: Colin Paice
Subject: performance issues
Usually puts to local queue are fast.
If the messages are non persistent, there is no logging. If messages are persistent then the ion could be slow. Check with other queues in the system, using persistent messages
If connection is local then there should be no network delays.
If app is network client, this could be the issue.
If you use other queues in same box do they have same issue?
Original Message:
Sent: 10/17/2023 8:35:00 AM
From: Brajendra Kumar
Subject: performance issues
We are experiencing performance issues in IBM ACE message processing using Queue Manager to read/write messages. ACE runs in AKS container and QM also runs in container. We get the desired performance when both ACE and QM are in same AKS cluster, but getting a reduced performance when ACE( in AKS) containing to QM running in ARO ( azure redhat) cluster or QM running in any platform other than AKS. We are suspecting some network issue, but Network team couldn't find any kind issue . Kindly let us know if any other things we need to look into.
- ACE Application Pod/Container(DNS : XXXXX.net) running in AKS connecting to MQ application Pod/Container(XXXXXX.net) running in AKS -> we are getting 66 tps
- ACE Application Pod/Container(DNS : XXXXXX.net) running in AKS connecting to MQ application Pod/Container(XXXXX.net) running in ARO -> we are getting 22 tpsMQ is Installed on Azure Redhat Openshift platform and ACE is on Azure Kubernetes Service (AKS).Message flow is getting/putting the message from a queue very slowly . So I need to change the mq.ini configuration or any network issue.
------------------------------
Brajendra Kumar
------------------------------