Kuassi Mensah Interview
summary
Oracle product management discusses scaling Oracle + Spark access by partitioning a single relational table so 100 Spark nodes can read in parallel, using logical partitioning so each node hits only its shard. Engineers add a driver-layer shortcut to avoid double type conversion, emitting raw SQL bytes into Scala to skip Java boxing. Also covers Oracle’s Hadoop/Spark integration and argues functional or microservice design reduces QA and debugging time.
AI-written summary · about
related talks
50mChallenges that everyone struggles while productionizing Apache Spark workloads
41mOptimising Apache Spark and SQL for Improved Performance
28mAn Apache Spark query's journey through the stack
21mChallenges of Spark Application coexisting with NoSQL databases
40mScaling with Apache Spark (or a lesson in unintended consequences)
15mOptimizing Query Performance in Big Data Systems
06mInterview
40mBest Practices and Optimization with the Spark Engine