Part of this page failed to reach you. Reload to try again — if it keeps happening, check your connection.
Tiarê Balbi Bonamini
What you learn today is gone tomorrow. I write to make it last — turning study notes and hard-won lessons into articulated posts about software, technology, and the craft of building things.
I've been writing software since 2008 — from Java monoliths to distributed systems, from São Paulo to Vancouver. I'm a Brazilian engineer living in Canada who believes the best way to learn is to build, and the best way to retain is to write.
I kept rebuilding the same SSE envelope every time I wrote an agent UI. AG-UI is the first serious attempt I have seen at standardising that envelope. In this post I strip the protocol down to its wire shape and rebuild a minimal Spring WebFlux endpoint that speaks it without an SDK.
I crashed a Two-Phase Commit coordinator on purpose in a small Kotlin simulation to measure how long participants stay locked when the coordinator vanishes between phases. The result is the part of 2PC the diagrams never show — and the reason I would model most cross-service writes as a saga instead.
Static RPS caps shed the wrong traffic. Concurrency is what saturates a service, not request rate. From my notes after reading the InfoQ piece on overload protection, Uber's January writeup on Cinnamon, and Netflix's QCon SF talk on service-level prioritized load shedding, here is why latency is the right control signal — and how a small priority taxonomy plus an adaptive concurrency limit keep the cheapest traffic shedding first.
I ran the same hot-key seat reservation workload two ways: Postgres with a version column and retries, and a single actor per seat. The actor design did not scale better — it moved the hard problem from concurrency control to routing and rebalance correctness, and that trade was the easier one to reason about under hot keys.