Close Encounters with the Third Kind of Database
summary
Argues that a converged OLTP+OLAP database is practical: MemSQL pairs an in-memory row engine that handles tens of thousands to hundreds of thousands (and in some tests millions) of events per second with a columnar archive for long-term storage, and ships a MemSQL Spark connector that moves data bi-directionally. Proven pipeline: Kafka → Spark Streaming → MemSQL at Pinterest ingesting tens of thousands/sec to drive real-time, city-level trending and search; a finance customer cuts analytics latency from hours down to minutes.
AI-written summary · about
related talks
06mThe Enterprise-Focused Startup
05mDriving the On-Demand Economy with Predictive Analytics
41mIntegrating Kafka with a Real-Time Database
05mTechnical convergence in enterprise database software
05mTeaching databases to learn in the world of AI
12mPower a Better Future
32mDiscover the Modern Way to Build Real-Time Analytics Solutions at Any Scale
43mArchitecting for Speed & Scale