Showing posts with label simulated event processing. Show all posts
Showing posts with label simulated event processing. Show all posts

Tuesday, July 10, 2012

Event processing in simulation mode - an Infosphere streams example

I have written in the past about event processing in simulation mode,  recently an article on the IBM developerWork describes an Infosphere Streams implementation of traffic simulation.  

The picture above is taken from the article, a vehicle generator generates simulated events of vehicles. Each round represents 1 second in real-time and cars are accelerating, stopping and behaving according to traffic rules.   This is an example where the stream processing is done not on real-life events, but on simulated events.    The reported benefit of using streams have been the large scale of this simulation (multiple vehicles, rounds of 1 second,  multiple streets).   Besides the scalability aspects, much of the simulated behavior may be expressed using operators on events.    Simulation has been one of the first areas that use some type of event processing, and using event processing for simulation closes a circle.  

Sunday, January 9, 2011

Event processing in simulation mode

You are probably familiar with the Microsoft flight simulator, that is seen in the picture, and indeed simulators became very common both in games and in training - such as pilot training which for many years heavily use simulators.   


In a recent blog James Taylor has written about the difficulty to validate new strategies, since strategic decision is translated to multiple tactical actions and possibly large number of operational actions. 






One of the possibilities to validate that the operational actions are consistent with the strategic planning is the use of simulation, where external events are simulated, and an event processing system is being used as a simulated business activity monitoring that checks the operational action results.   This requires the ability to codify the strategic decisions as  a collection of KPI (key performance indicators), which is typically doable.
Building the simulated BAM is more challenging,  first it requires some enhancements on event processing technology (on which I'll write in a later post), then it require to codify the operational actions and their effect on KPIs into the system.    I view event processing systems with sophisticated capabilities as a kernel for simulation and validation systems, not only in a corporate strategy level, but also in larger societies, simulating social processes,  or the impact of government policies.    One of the directions to pursue further - more about this direction later.