Cloudflare Redesigns D1 Database Control Plane
The redesign lets Cloudflare's serverless SQL database support millions of databases by splitting metadata, analytics, and routing across separate systems.
Updated on Nov. 8, 2026 in Data Centers
Cloudflare redesigned the control plane of its D1 serverless SQL database to support millions of databases. The new architecture keeps query processing at the edge, separate from centralized management operations.
Why it matters
Daily active D1 databases grew 18-fold over the past year, so the platform had to scale its management layer without creating bottlenecks. Separating query processing from control operations is how the system is designed to handle that growth.
Daily active D1 databases grew 18-fold over the past year, and the control plane reconciles each database with Postgres through 30-second heartbeats with per-database jitter. The article does not report performance results at the full scale of millions of databases.
The players
Cloudflare
Cloudflare is a global internet infrastructure company that operates content delivery, security, and serverless computing services.
The details
The control plane uses Postgres for transactional metadata, ClickHouse for analytics, and Workers KV for globally replicated routing data. Cloudflare routes D1 requests from an Edge API Worker through an Eyeball Worker directly to the relevant D1 Durable Object.
Timeline
2022: Cloudflare D1 was announced.
Past year: Daily active databases grew 18-fold.
November 8, 2026: The article on the control plane redesign was published.
The Tech Race
The redesign builds on Cloudflare D1, which the company announced in 2022 as part of its serverless SQL offering. It positions Cloudflare to compete on large-scale, edge-distributed database management.
Developers using D1 can handle growing workloads across many databases without managing the underlying infrastructure. Free-tier limits of 5 GB storage, 5 million rows read per day, and 100,000 rows written per day still apply to smaller projects.
The takeaway
Scaling a serverless database to millions of instances depends on separating query processing from centralized management. Developers building on distributed platforms should understand where their requests are actually handled.
Further reading
For more on how infrastructure like this is built and operated, see Data Centers.
Source note: This article includes information reported by InfoQ.
