Kickoff Day 3
summary
Claims that one engine for ML streaming graph SQL (Spark) can run on top of existing clusters and that Hadoop 3.0: resource mediation + smart storage will evolve Yarn into workload manager and add RAM and large‑flash tiers to HDFS. Databricks can run Spark self‑contained, Hadoop vendors will integrate Spark, and cloud providers push end‑to‑end data stacks as a service. Docker has dominant momentum while CoreOS risks overexpansion and EMC is predicted to reinvent itself with VMware central.
AI-written summary · about
