Why Distributed Real-Time Stream Processing must have a Persistent Local/Embedded Store like RocksDBSometime back in year 2013, Facebook announced to open source their embedded storage: RocksDB Initially, I could not realise the significance of that and thought its just another solution for their set of problems. Lately, started noticing RocksDB being mentioned a lot frequent in recent times during discussions of distributed stream processing frameworks like Apache Flink, Kafka Streams, Samza ,
