Unlocking the next wave of applications with Stream Processing

on YouTube ↗49m

" summary

New SDK exposes addressable virtual functions that hold per-entity state, run on Flink, and serialize state with Protobuf for schema evolution. Messages route via routers from Kafka ingresses and return to Kafka egresses; the runtime maps functions to a Flink job with a function operator and a co-located feedback operator so inter-function messages are checkpointed for exactly-once state checkpoints. Demo: a ride‑sharing app with driver/passenger entities works end‑to‑end. Current limits: processing‑time semantics, no windowing yet, infinite feedback loops can starve the job.

AI-written summary · about

" chapters (28)

1. 0:00Intro
2. 0:20Overview
3. 1:04Decay
4. 4:12Summary
5. 5:19Stateful applications
6. 5:51Functions
7. 6:23Address
8. 9:25Execution Model
9. 13:13State
10. 14:04Evolution
11. 15:21Ingress
12. 16:34Routing
13. 21:06Jail Cell
14. 26:41Visualization
15. 26:53Under the hood
16. 30:52The function operator
17. 32:46How to deploy
18. 33:55Deploying stateful function modules
19. 34:58Launching stateful function modules with Flink
20. 35:27How do we get started
21. 36:05State Machine
22. 37:05Metrics
23. 38:28Integration
24. 39:14Feedback Loop
25. 40:58Messages in Feedback Loop
26. 42:21State Performance
27. 43:40Updating Stateful Functions
28. 44:57Queryable States