RisingWave is a distributed system with the following components:
📄️ Internal State Management
Stream processing systems continuously process data streams. The core design revolves around managing the internal states during the continuous computation process. RisingWave uses remote object storage (such as S3) for persistent state storage and caches it on the local memory and disk of computation nodes. The internal state storage in RisingWave is referred to as Hummock, which is structured as an LSM tree. The compaction process within Hummock continuously sorts data and persists it to remote object storage.