GSoC 2024...
summary
PGWatch v3 adds a new RPC sync layer that makes sinks pluggable and sync-agnostic, replacing the need to hardcode many formats. A small Go receiver interface exposes update_measurements and sync_metric hooks; examples include CSV, Kafka, ClickHouse, S3 and a Kafka receiver that maps databases to topics. The demo wires a TinyLlama model via Ollama, runs with batch_size=1, writes LLM-generated insights back into Postgres and exposes them over a REST API, and the implementation works end-to-end.
AI-written summary · about
chapters (6)
- 0:00Introduction to GSoC
- 6:17What is pgwatch
- 10:47Need for remote syncs
- 17:25Developing custom syncs
- 23:29Live demo with LLM
- 30:13Q&A and closing
related talks
27mEvolution of serialization formats and usage in Go
55mHow Go powered our elastic search
43mWhen it all GOes right
32mA Journey to Postgres Productivity with Go
26mTime to Go with PostgreSQL
47mA Journey to Postgres Productivity with Go
30mCall me back, Postgres
37mLet's build our own message streaming platform