Saturday, October 23, 2010

On IBM Websphere Decision Server

IBM recently announced "Websphere Decision Server".  This announcement states that this offering combines business event processing with business rules management system to accelerate decision making.


There are several ways in which event processing and business rules interact, some of them are: an event, derived by pattern detection, triggers a rule, which makes it event-driven decision; the other direction is also valid:  an execution of rules brings into decision and this decision can be reported as an event and may influence other decisions.  Triggering decisions is one of the major uses of event processing (of course it is not the only one,  e.g. event processing can be used for diagnosis or information dissemination), and it is a component in the automated decision making process, but these components mainly exist within islands (EP, business rules, optimization software and other means of decision support systems).    One of the areas we are working on (we had a short report on it within the "fast abstract" session of DEBS 2010, as we were not in a position yet to deliver a full paper) is a unified conceptual model of event processing and business rules, where both are generalized as decision agents.  this is still in the research phases, and not part of the product, nevertheless this announced offering provides step forward in achieving such an integration.   More synergies in the decision space are expected.     

Tuesday, October 19, 2010

Teaching an event processing course again

Yesterday I have started teaching again, this time in the Bloomfield building at the Technion, seen in this picture.   Sometimes people ask me why do I continue teaching,  it is probably not because I am bored, on the contrary,  I am quite busy;  it is probably not because it is a well-paid job,  actually they pay peanuts.  
I think that I continue to do it, since I like the interaction with students, and it is also means to evangelize "event processing" to the younger generation;  it also provides exposure to graduate students to work with, as over the years I have supervised (or co-supervised)  6 PhD dissertations and 21 MSc thesis, if my count is right.   It is also source for recruiting.   Quite a lot of my colleagues in IBM have been my own students (which I supervised), or my teaching assistants, or students that took courses with me.    


The way my courses are designed revolve among the following principles:


For historical reasons the course is called "database engineering" and always there is a student who expects to see classical database course,  for these students I answer that today in any database conference, data stream management is one of the emerging areas, and I am giving a flavor of it (my approach is not database centric,thought), I also show students some analysts testimonies that event processing is an emerging markets in general.   The second principle is deep dive -- the course is going more on depth then on breadth, the students should gain deep understanding in what event processing is. The last principle is retain over time, my own experience is that to retain over time students have to experience themselves, so they'll get implementation project, I have not yet decided what the projects will be, gave myself time until next week.
I am using the EPIA book as a text book,  but this time will not focus on the FFD example.   
I might create teaching materials for this course so that others will be able to reuse it.