About This Simulation

This interactive simulation demonstrates how Kafka Streams manages state stores, including the restoration process, RocksDB caching behavior, and the threading model for state management. The visualization shows the complete lifecycle of state stores from restoration to active processing.

Key Features

• State Store Restoration: Watch how state stores are restored from changelog topics in the StateUpdaterThread before transitioning to active processing in the StreamThread
• RocksDB Cache Behavior: Observe cache hits and cache misses during processing, showing how RocksDB serves reads from its in-memory cache or fetches from disk
• Volume Persistence: See how volumes persist RocksDB state data on disk, enabling fast restarts when valid data exists. Click the trash icon to invalidate the data on disk and trigger a restart with restoration
• Changelog Synchronization: Follow how state changes are written to both the RocksDB volume (for local durability) and the changelog topic (for recovery)
• Multi-Partition Support: Switch between 1 and 2 partitions to see how multiple tasks operate independently with isolated state stores
• Store Type Comparison: Toggle between RocksDB (persistent with caching) and in-memory stores to understand the trade-offs
• Task Restart Behavior: See the impact of restarting a task with either a RocksDB or an in-memory state store. A RocksDB-backed state store can re-use its volume (if still valid), while an in-memory state store always has to restore from the changelog