Peter Boncz | Hadoop Summit 2014
summary
Claims a working analytical engine that outperforms legacy SQL-on-Hadoop options, citing fastest database engine results on single-server TPC-H workloads and Vectorwise lineage. It ships as Actian Vortex, integrates natively with HDFS and YARN, uses compressed columnar layout and a patented positional Delta trees structure to support trickle updates without shutting down HDFS, and targets production SQL workloads rather than wrapped legacy engines. An evaluation release is scheduled at the end of the month.
AI-written summary · about
chapters (22)
- 0:00Introduction to Hadoop Summit
- 0:49Peter Boncz introduction
- 1:10History of Vectorwise
- 2:38MonetDB and column stores
- 3:20Database technology spin-offs
- 4:03Big data and databases
- 4:48Actian Vector and Hadoop
- 6:12Vectorwise customer applications
- 6:52Vortex as native Hadoop
- 7:58YARN integration benefits
- 9:03Latency challenges in Hadoop
- 9:46Legacy SQL-on-Hadoop issues
- 10:51Positional Delta Trees updates
- 11:36Future-proofing database architecture
- 12:41Maturity of SQL systems
- 13:02Democratization of cluster computing
- 14:48End of silicon scaling trend
- 15:51Challenges in parallel programming
- 16:57Abstraction in parallel tools
- 17:59Vortex roadmap and elasticity
- 19:07YARN resource management
- 20:33Resources and contact information
