Cloudflare

Our billing pipeline was suddenly slow. The culprit was a hidden bottleneck in ClickHouse

A partitioning change to a petabyte-scale ClickHouse cluster caused billing pipeline jobs to stall without obvious error signals in standard metrics.

databases observability
4 min
Cloudflare

Agents that remember: introducing Agent Memory

AI agents lack persistent memory mechanisms to retain context, learn from interactions, and improve decision-making over time.

storage-systems ml-systems
3 min
Cloudflare

Artifacts: versioned storage that speaks Git

Providing agents, developers, and automations with scalable, Git-compatible versioned storage that can handle tens of millions of repositories without forcing them to manage infrastructure.

storage-systems api-design
4 min
Cloudflare

Unweight: how we compressed an LLM 22% without sacrificing quality

GPU memory bandwidth constraints were limiting LLM inference efficiency across Cloudflare's distributed edge network, requiring optimization to deliver faster and cheaper inference.

ml-systems distributed-systems
4 min
Cloudflare

A one-line Kubernetes fix that saved 600 hours a year

Cloudflare's Atlantis instance took 30 minutes to restart due to a Kubernetes volume permission bottleneck.

observability storage-systems
4 min