siddontang

Love this: DuckDB + SlateDB + S3. A real, writable database. One writer, many readers. No always-on server. Beautifully simple architecture. 😂 https://t.co/OwOqKcWkNq

Another RocksDB replacement. Nice. RocksDB is fast enough for many workloads. Its moat is years of production experience and hard-won reliability. With AI, I’d rather build a specialized engine than bet on another young one. We had to do that for TiDB X because of object storage. Writing the engi…

My latest experiment: DuckDB × TiDB File System. A laptop can commit a write, close the database, and a worker on another machine can reopen it and continue. The native database and WAL stay on shared storage. That is the foundation for a DuckDB service: shared database state with replaceable comp…

I am a bit of a SlateDB fan. A few friends recently asked me what it is, so I made this short intro. I hope it helps more people learn about SlateDB. Thank you, @criccomini , for building something so cool! https://t.co/9i7jbRE8Hn

RocksDB knows which SSTs will die young. SSDs don’t. 😅 A new paper uses NVMe FDP to tell the drive. In one test, device WAF fell 1.91× → 1.28× (~33%) with ~25 lines of RocksDB changes. No new compaction algorithm. Just a hint. Amazing idea. https://t.co/NMFRJYnc4d