When the gateway fell over
What I learned by building five rate-limiting algorithms and putting them under the same load.
How software works. How we work together. Two sides of what I write about, with room for code, questions and people.
Implementations, architecture and experiments. Decisions you can examine in code.
What I learned by building five rate-limiting algorithms and putting them under the same load.
The same number-porting workflow built with queues and virtual actors. What each approach makes easier, and what it asks you to own.
A grain from end to end: state, third-party callbacks, correlation and timeouts.
Replacing a direct call with streams, and examining what fan-out changes.
Moving state, reminders and cluster membership from memory to Postgres.
Atomic operations across grains, and the trade-off between transactions and compensation.
Following a request across the cluster with .NET Aspire and OpenTelemetry.
A critical look at actors and microservices: infrastructure, complexity and the shape of the problem.
The other angle of the closing: what the experience at Halo scale tells us about Orleans.
Technology, teams and people. Questions about how we think, relate and build together.
Using AI to build verifiable tools instead of trusting its conclusions. A reflection on evidence, engineering judgment and checks that outlive a conversation.
Self-awareness, the distance between how we appear and how we think, and the assumptions we bring to our teams.