What do you like best?
I used ODM to detect changes in a flume streaming application. My top pro was its ability to only execute rules where it had all the data to make a decision. That allowed me to decouple many of the streaming nuances away from the logic that determined the outcome for this event detector. The time spent on memory and speed of processing was very beneficial in this design.
What do you dislike?
ODM is a bit clunky, its proprietary, and the documentation was not great. Coming from an Application Developer background across big data, salesforce, and open source libraries ODM can be a bit of a shock on the documentation. Either its not well publicized or it wasn't great for what was given to me. World class documentation would go along ways. "Self-Service" documentation.
Recommendations to others considering the product
Make it more available to entry level developers. Documentation, speed of implementation. Those were really big issues for my prototyping in this space.
What business problems are you solving with the product? What benefits have you realized?
I was solving for Event Detection within a streaming application. Rule Engines can be embedded in stream processing to achieve this design. IBM's Decision Server Insights is also a great fit for this need.