Solving the Big Data Challenge of Sensor Data

IBM Webcast June-25-2013
Bernie Spang Director, Strategy and Marketing, IBM Database Software and Systems
Kevin Brown Chief Architect, IBM Information Management

Sensor Data Brings New Applications Across Industries
IT Challenges of Sensor Data Applications

• Sensors output data repeatedly over specified time intervals – Data system must be able to receive and process data at extremely high input velocities
• Data output is time-sequenced resulting in a different structure – Data system must be able to store large volumes of data in a scalable fashion and allow near real-time access to it
• Data naturally includes geospatial information – Data system must be able to efficiently store, correlate, and analyze data with repect to time and location
• There is a higher need to deal with communication failure – Data system must be extremely reliable and continue to operate in case of hardware failures
• Queries may need to access sensor data along with relational data – Data system must provide near real-time analysis of the collected data

Time Series Data is BIG DATA - Large amounts of data cause problems in 2 areas:

It’s expensive and difficult to manage

• Massive storage space required
• Sophisticated partitioning schemes needed
• Frequent reorganization leads to downtime

Poor query/analytic performance affects quality of service

• Compliance reports must be completed before the end of each day
• Customer portal queries must be handled in a timely manner
• Customer billing must be completed each day

Accessing TimeSeries Data with Informix

• Access through standard tabular view – Makes TimeSeries look like a standard relational table
• SQL Interface – 100+ functions
• Extensive library of API functions – Write your own functions in Stored Procedure Language (SPL), “C”, Java

 Solving the Big Data Challenge of Sensor Data Webcast (2013-06-25)