BLAST - A Big Data Platform for Analytics on Spatio-Temporal Data - overview
Numerous smarter planet applications (weather and environment modeling, smarter transportation, security, network analytics etc.) generate/consume/analyze scientific data with spatio-temporal characteristics. IBM is currently engaged with various customers (e.g., Meteorological Department of Brunei, National Environment Agency of Singapore, Bharti etc) which are generating and analyzing such scientific spatio-temporal data. For example, Meteorological Department of Brunei generates a weather data which consists of 300 weather parameters at one lattitude-longitude and at a given timestamp. Combined across multiple locations and timestamps, this data runs into TBs for each month. Such data needs to be analyzed in a number of ways e,g,, to retrieve the data satisfying certain conditions, analytics routines like Interpolation, Aggregation, Iso-surface Finding, Pattern-search, Sampling, Histogram representation etc. Managing and analyzing such Big spatio-temporal data on a map-reduce platform throws a number of system challenges - both research and implementation.
However, even as these data sets continue to grow in volume, there is little to no support on Big Data platforms to support these applications. Toward this, we at IRL, are working toward BLAST - A Big-Data Platform for Analytics on Spatio-Temporal Data. The goals of BLAST are - i) A comprehensive spatio-temporal data management and analysis stack built on top of IBM’s BigInsights platform, ii) Support for mining, model-building, and other forms of spatio-temporal analytics on large volumes of spatio-temporal data, iii) Capabilities to be designed (and validated) using real use-cases from multiple ongoing projects within and outside IRL (NEA, Brunei, Network Analytics for Bharti, etc.).