ClickHouse parallel replicas: 100B-row GROUP BY in under half a second
Abhishekcur · x · 2026-10-18
ClickHouse Cloud introduced parallel replicas (an OSS feature, now in beta on Cloud and heading to GA): a single query, including complex GROUP BYs, can automatically scale across thousands of cores.
The official demo: a GROUP BY over 100 billion rows completed in 0.414 s, with throughput of 241.8 B rows/s (1.45 TB/s) — no pre-aggregation or data reshuffling required. It works essentially as virtual sharding on demand: add nodes and queries get faster, with no data movement. ClickHouse treats one 90-core machine and 100 machines with 9,000 cores identically.
More from Infra
- Three receipts in 14 days: Anthropic's $15B SpaceX compute deal fuels the 'last window' thesis — elonmusk · 2026-10-18
- Cloudflare Now Scores Verified Bots on robots.txt Compliance and 429 Response — shashib · 2026-10-18
- AI-generated kernels were mocked 2 years ago as 'just a while loop' — the world has changed — bingxu_ · 2026-10-18
- ClickHouse aggregates 100B rows in 0.4s with parallel replicas auto-scaling — Abhishekcur · 2026-10-18
- Auction-based LLM routing paper finds fixed token pricing wastes 10%-71% in costs — _onionesque · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18