Interview with – Andrey Fomichev
summary
A distributed transactional database grows from internal Yandex scale work into an open source product that already runs in Yandex.Cloud and supports on‑prem installs, replication and a practical consensus that uses a master-per-entity plus quorum design instead of pure Raft/Paxos. Claims include running clusters larger than early Cassandra installations, performing a live 100-shard migration between data centers, and offering cloud and paid support/managed services; the system proves usable in production and continues active feature and performance development.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:48How came to databases
- 10:19Career in projects and Yandex
- 17:51Storage systems evolution at Yandex
- 24:52New distributed system development
- 34:09From internal to open service
- 38:09Open source choice and future plans
- 44:10Database development trends
related talks
42mReliability of High‑Load C++ Applications in Yandex.Market
06mGlobal Scale Multi-Cloud, Multi-CDN Load Balancing with C++
38mScaling Financial Transaction using 0MQ and JSON
54mlibfpta — Outrunning SQLite and Tarantool
55m “High Performance C++ Concurrent Transactional Data Structures"
34mCrimson Project Update
01h 22mBeyond Sequential Consistency - Leveraging Atomics for Fun & Profit
27mRebuilding a Faster OSD with Future