Meta changed one character of code and freed up 15,000 servers a year — thanks to eBPF

blaizedsouza · x · 2026-09-06

• Meta's Strobelight continuously profiles every production host at near-zero overhead; that visibility surfaced a one-character code fix saving capacity of 15,000 servers per year.
• How eBPF works: tiny sandboxed programs loaded into the running kernel — no patch, no reboot, no downtime; a verifier checks every instruction so the kernel cannot panic.
• Use cases: Cloudflare's XDP drops 10M+ packets per second per core inside the NIC driver, mitigating multi-terabit DDoS floods; AWS ships eBPF-based Cilium as EKS's default network layer, pushing 30-40% more throughput.
• The author also open-sourced pktz, an eBPF-powered live traffic monitor that accounts every byte per process and per connection, no packet sampling required.

Original post →

More from Infra

• Foxconn Sees Q3 Beating Expectations on Growing AI Server Demand — firstadopter · 2026-09-06
• ComfyUI reloads models from disk on every run; --disable-dynamic-vram is the workaround — Silver-Spot-2763 · 2026-09-06
• Providers offering DeepSeek V4 Pro 0813 differ by fractional cents in pricing — DanielLockyer · 2026-09-06
• Google CTO: Taiwan supply chain now co-designs chips and racks, no longer buyer-supplier — pstAsiatech · 2026-09-06
• The Biggest Bottleneck in the GPU Industry Isn't Chips or Power — It's Credit — rickasaurus · 2026-09-06
• Oracle expands HPE Juniper deal to network its gigawatt-scale AI data centers — shashib · 2026-09-06

Infra channel →