ClickHouse aggregates 100B rows in 0.4s with parallel replicas auto-scaling
Abhishekcur · x · 2026-10-18
ClickHouse engineer Tom Schreiber details the new Cloud capability parallel replicas (beta, GA soon): a single query, even complex GROUP BYs, automatically scales across thousands of cores — like on-demand virtual sharding with no data movement or pre-aggregation.
Key numbers:
- 100 billion rows aggregated in 0.414 seconds
- 241.8 B rows/s throughput (1.45 TB/s)
- A 90-core machine and 100 machines with 9,000 cores behave identically to the query — just add nodes
The 34-minute post walks through why GROUP BY is the heart of analytics, how ClickHouse makes it fast, vertical vs horizontal scaling, how work is distributed across replicas, scaling limits and safeguards, and the measurement methodology.
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
- 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
- SGLang Summit sponsors jump from 11 to 36 in one week — ying11231 · 2026-10-18