FM3 — The Queue That Never Stops Growing
FM3, rate limiting, and the defence-in-depth assumption that turns a spike into an outage
Articles on system design, failure modes, architecture tradeoffs, and engineering leadership.
FM3, rate limiting, and the defence-in-depth assumption that turns a spike into an outage
The four-step method: find the bottleneck, name the data structure, name the algorithm, name the complexity
Metcalfe’s Law, power-law nodes, and the architecture cost that arrives before the value
Traverse from category to structure to risk — not from symptom to fix.
Every interface hides complexity. A good one chooses, narrowly and correctly, what to expose.
It looks like a CRUD app and behaves like a distributed system — the two tradeoffs that decide it.
Why a B-tree is just a binary search tree redesigned for the cost of a single disk read.
Onboarding cost, quadratic communication overhead, and a contributor who breaks even after the deadline.
Throughput and latency pull against each other past saturation — and the mean latency hides the system you actually shipped.
Nine lenses for seeing, building, and diagnosing systems — and the dependency chain between them.
The mechanism that makes complex systems manageable — and why they all leak eventually.
The TTL, the token, and the three failure modes that cause silent data corruption.
Context, container, component, data flow, and deployment — each answers a different question.