Please share with the community what you think needs improvement with IBM Integration Bus.
What are its weaknesses? What would you like to see changed in a future version?
I would like for them to make the training much easier. Once you work with a sphere of people that can integrate the solution, then it's stable. But just to educate people, it's not that easy to do. It's not easy content to teach people.
IBM could improve its connectivity. The solution needs to be better integrated with the cloud version of the solution. The solution is constrained by the environment - whether it's on-premises or the cloud. The cloud seems to be more adaptable.
Currently, we have some custom solutions for logs; it would be nice if these solutions could be provided out of the box. The solution doesn't have a standard testing framework that can be used for integration performance and other tests. The user interface could be improved in a future release.
This product uses the PVU (Processor Value Unit) license model from IBM, and it is something that should be improved. It requires you to install monitoring software that ensures you are only using the number of CPUs that you have paid for. This license monitoring tool is very complicated.
I like the IBM Integration Bus and I hope that it will change in micro-service architecture. My understanding is that it will change to be less connected, and less depends on the operating system or the hardware resources. I would like to be able to run and install this solution on different platforms and using containers and using modern micro-service and cloud environments.
This solution would benefit from improvements to the configuration interface. It is hard to understand, and one small change can have a huge impact. For example, if you say Yes instead of No in one of the configuration settings, or Transactional instead of Non-transactional, then the whole meaning changes and it is difficult to track down the problem. This is the reason that many of our projects are progressing slowly. We just don't know what is going to happen with different parameter settings. It makes it very difficult to be creative. The only other difficult part is that IBM adds its own meta-data, in addition to the normal, generic XML data, into the tree. It's hard for us to understand how to navigate the tree and pick what we want or figure out where our own application data lies, because of the additional IBM specific data. We understand that they do things this way in order to reduce the programming, but it's more of a learning curve.
It would be better with more API management features. More Micro-service and container based support. IBM is already working on it on Version 11, but it still needs improvement. Also IIB have cloud version ,it is doesn't have all features of the On premise version , and needs more improvement .
The resources about IBM are hard to find, and it is not enough material. Finding people who know this product is problematic.
IBM does not support orchestration, which is how they designed it, and other BPM tools in the market support orchestration. IIB wasn't designed for short and small transitions which are preferred to be stateless. If they merged the BPM capability into this product, then it would be a better solution.
* My biggest concern is its MQ dependency, which are still not 100% independent. E.g., in the case of aggregation flows, IIB needs a complete overhaul for aggregation implementation to achieve orchestration implementation. * Development toolkit (based on Eclipse) should be improved in terms of responsiveness. * It lacks unit testing framework similar to JUnit for Java or MUnit for Mule ESB. * IIB run time and installation still need to shed more weight and become lightweight for to become full Docker applications.