A big data technology designed to enable real-time updates to analytic databases has earned a U.S. patent. The approach was developed by Dutch computer scientists, and the technology was eventually acquired by Actian Corp.
Patent No. 9,892,148, “Methods of operating a column-store database engine utilizing a positional delta tree update system,” was awarded Tuesday (Feb. 13) to the data management vendor based in Palo Alto, Calif. The data structure is designed to boost analytical performance by accelerating database updates. The delta tree approach is touted as overcoming the batch loading and reporting schemes used in Hadoop-based systems that are tailored to scanning only existing records.
A positional delta tree is implemented in database engine memory and linked via an update data transfer path between the database engine and data tables. The data structure stores “differential update data values” in defined positions to compare those values with a stable data table.
Actian has deployed the patent technology in its Vector analytics database. The company claims its patented approach addresses the architectural limitations of the batch process associated with Hadoop-based systems to provide real-time analytical computing. Those analytical workloads must be integrated with operational platforms to help reduce performance shortfalls stemming from database updates, industry observers note.
Actian acquired the positional delta tree technology from the Dutch research group CWI, then added the data update approach to its flagship Vector database.
The Dutch researchers filed a U.S. patent application in 2012. According to the patent awarded this week, “the database engine is responsive to database requests received with respect to database tables stored by persistent data stores accessible by [the] database engine.