Saturday, January 1, 2011

On interfaces and event processing


Interfaces is one of the main next frontiers for event processing systems.  In the picture you can see an illustration of brain-computer interface which is the most advanced interface that is mentioned in a recent human-computer interface survey,  this survey starts from command line, moves through mouse and keyboard and gets to gesture detection, and up to the mostly futuristic brain interface.  My daughter Daphna got for her recent birthday XBOX machine with Kinect, which is capable of voice and gesture recognition and provide advanced game experience.   


Getting to event processing, the human computer interface is crucial in order to raise the abstraction level and enable to extend the type of developers and users, currently the development interfaces are geared mostly towards programmers, and are in essence close in nature to the implementation abstractions, a separation between the implementation abstractions and the development abstractions is one of the major challenges, in general in software, but an important one in event processing, since realizing the event processing potential entails the ability to make application development accessible  to other communities such as the spreadsheet programming community (e.g. business analysts).   I have written in the past about this requirement, and will probably deal with it more in 2011. 


Talking about interfaces, there is another type of interfaces the application programming interfaces (API), this is important since one of the major utilization of event processing is as embedded capabilities inside applications/middleware/solutions.   Thus, APIs that are appropriate for multiple uses are gaining importance for the interoperability and  for making EP  easily embeddable.   API design has become an emerging area, a blog posting by Shanley Kane from Apigee (and API management provider) provides trends and predictions for APIs.  It is interesting to note that the first prediction is entitled "APIs go real-time big-time" (in the source all words are capitalized, but I have adopted Manning's writing style while writing the EPIA book which is against over-capitalizing).   This prediction talks about the popularity for APIs of push oriented pub/sub systems.   This is elementary event processing,  APIs for more sophisticated event processing is the next step.
Standard APIs for EP will enable plugging EP components that are doing aggregation, translation and pattern matching components as part of an "event-driven Web" (AKA event-processing fabric) that was declared as the grand challenge in the EP Dagstuhl seminar.  I'll write more about this grand challenge soon.


In any event, both human computer interfaces and application programming interfaces are major part of the work needed towards the next generation of event processing systems. 



No comments: