Clive, one of our lead Solution Architects with a lifetime of delivering IBM FileNet solutions, blogs about his Journey to the Cloud with deployments of IBM Cloud Pak for Business Automation.
To catch up on his previous blogs – here are parts 1, part 2, part 3, part 4 , part 5 and part 6.
🎺The Last Post
It’s been a while, but with this update, my journey is complete.
Almost three years ago, the world was hit with the Covid pandemic, and at the same time, I started a journey with IBM Cloud Pak for Automation products and different cloud technologies. For those that have read the last six posts, you’ll know this was due to a project where the customer wanted to deploy IBM Cloud Pak for Business Automation. This project is now live, and a happy customer is using a full production system with over seventy million documents.
I’m not going to cover the project specifics, but I will discuss some observations and conclusions I’ve made over the three years.
Private Cloud hosting is complex to set up, specifically on the networking side, but once set up and configured, upgrades are simplified. RedHat has improved the setup procedures of Openshift, which have simplified the final install of clusters, but there is still much pre-work to be completed on the networking side to have a smooth deployment. The thing I’ve used most in the last year or so is the assisted installer. This is wizard-based, creates an ISO file to download and boot preconfigured machines (Virtual or bare metal), and with a small amount of information about the target environment, can create multi or single-node clusters.
I have also been successfully using Openshift Local on my mac, but this has 64GB RAM and 6 core i7 processor giving 12 threads.
Cloud Pak deployment has all the same pre-reqs as the more traditional deployments, databases, directory services for user authentication and authorisation and file storage it is just a different skill set needed. Gone are the times of using the IBM FileNet Configuration Manager tool and Websphere Application Server Console. Now it is making sure there is the correct number of spaces in YAML files.
Finally, the big switch to administering a system once active is remembering that some of the configuration has to be made by changing the deployment YAML rather than traditional admin tasks. The biggest one of these is adding an object store. Adding one in ACCE will result in the operator removing it; the only way around this is stopping the operator execution, but then there is no way to upgrade.
I have noticed, but not tested, IBM has now added a BAW deployment which I believe is a step between the core FNCM and IBM Cloud Pak for Business Automation. If it is anything like the change in complexity and system resource requirements that I discovered moving from the Cloud Pak to FNCM, I would strongly recommend anyone looking to deploy using this method, just make sure it has all the components needed for the project.
That’s it from me, signing off…
Good luck all.