Chris Phillips’ Blog - API, Integration and Governance

This was originally published on 3rd October 2022 on IBM Community. Republished with permission from author

It is a common mistake to simply add more resources to a system to get more performance. For example, one might enlarge the thread pools and database connection pools, assuming more threads can do more work and therefore process more messages. This can be true up to a point, but beyond that it may result in no performance increase, or even significantly reduced performance. This article attempts to describe a situation where these two things can happen.

Having a good set of in-house API Governance rules allow your developers to validate that their APIs meet their enterprise standards. Other rules can be used to ensure serviceability considerations are made to support the API or can highlight where anti patterns are being used.

I have been challenged by API Connect’s Product Management to come up with some sensible governance rules. I expect this article to grow from time to time as new ideas are formed.

I don’t post that much about my personal life on this blog. I also haven’t posted much at all for the last few months. This will be one of the few posts I talk about my own life, and I wanted to share my experience encase anyone else is going through something similar.

In July 2023 my wife and I discovered we going to have another child. We already have two amazing daughters (11 and 13 at time of writing).

Canary deployments are a way to roll out changes to a service in a controlled manner, by gradually introducing the changes to a small group of users before rolling them out to the entire user base. API Connect can provide the functionality to different versions of a downstream service using a simple piece of gateway script.

When deploying API Connect in OpenShift you can deploy the API Connect Cluster Object. Sometimes this is referred to as the TopLevelCR. This will deploy a management, analytics, portal, and gateway. If you are using CP4I it will also auto configure it as well.

However, there are times that all subcomponents are not required.

A few years ago I was playing around with jmeter-on-ocp. https://github.com/ChrisPhillips-cminion/jmeter-on-ocp. Please note this is my fork, I am unsure what is happening to the original but credit must go to https://github.com/alvarolop for the start of this.

As an API Connect platform management team being able to confirm the API Manager and Gateway are working as expected is critical. The API Manager and the Gateway both have key endpoints that should be monitored to determine platform health.

Geraint is one of the Senior Engineers in the API Connect Analytics Squad. He has written a great article on how Rollover and Retention is handled in APIC V10.0.5.

https://community.ibm.com/community/user/integration/blogs/geraint-jones/2023/09/07/analytics-rollover-and-retention-in-v1005

This is a great article by Rashid on handling API errors! https://rashidaljohani.github.io/API-error-handling-why-should-you-care/