Featured Article

Achieving Sub-10ms Latency: A Deep Dive into Anycast Routing

When milliseconds define user experience, the architecture of your routing layer becomes your most critical engineering decision. We rebuilt our anycast topology from scratch — here's every lesson we learned.

JK
James Kim · May 8, 2026 · 12 min read
🌐
Latest Posts
🔐
Security

TLS 1.3 in Production: Benchmarks, Pitfalls, and Rollout Strategy

We migrated all 47 edge nodes to TLS 1.3 over six weeks. The 0-RTT resumption gains were real — so were the unexpected replay attack vectors we had to mitigate.

⚙️
Infrastructure

Zero-Downtime Deployments at 400Gbps: How We Do Rolling Updates

Draining connections at line rate without dropping packets requires more than just a graceful shutdown signal. We open-source the tooling that makes our zero-downtime deploys possible.

Performance

QUIC vs TCP: Real-World Performance on Lossy Mobile Networks

HTTP/3 over QUIC promises better performance on unreliable connections. We ran 90 days of A/B tests across six mobile carriers in four countries. The results surprised us.

🗺️
Networking

BGP Communities as a Traffic Engineering Primitive

Most teams use BGP communities only for basic blackholing. We use them as a first-class traffic engineering primitive — steering flows across 12 upstream providers in real time.

🛡️
Security

DDoS Mitigation at Layer 3/4: Lessons from a 1.8Tbps Attack

Last November we absorbed the largest volumetric attack in our history. This post is an honest postmortem of what worked, what failed, and the architectural changes we made afterward.

📊
Infrastructure

IPv6-Only Internal Networks: A Migration Guide for 2026

We completed a full IPv6-only migration for our internal backbone in Q1 2026. Dual-stack is a crutch — here's how to rip it off cleanly without breaking your monitoring stack.

Stay in the loop

Get new articles on networking and infrastructure delivered to your inbox. No spam, unsubscribe anytime.