With successful application of Big Data principles to many batch oriented processing problems, Nokia is applying those principles to stream-based traffic data using a distributed actor model. This paradigm heavily emphasizes “separation of concerns” by allowing the developer to concentrate on writing business logic without considerations for infrastructural and other cross-cutting concerns like messaging and threading while allowing that code to be applied to a massive amount of real time data on a widely distributed cluster.
Most statistical processing problems can use this paradigm, which involves decomposing the problem into: 1) simple discrete “actors” 2) an actors addressing scheme. We will walk through a use case which forms the precursor to many of the algorithms we use to create traffic products and see how they are implemented on Nokia’s new open-source real-time map-reduce system, Dempsy (http://dempsy.github.com/Dempsy/).
SPEAKER BIO: Jim Carroll, Navteq Fellow
Jim Carroll is a Navteq Fellow at Nokia, Location and Commerce. He is the Lead Architect for the Traffic Products and is currently the project lead for Dempsy, Nokia’s Stream-based Real-time Map-Reduce system that was just open sourced. Jim earned a Master’s Degree from Villanova University.
Dempsy PPT file