Production lessons
The green dashboard lied to us
Every metric was within threshold. The service was down for half the users. Availability and health are not the same thing.
Notes from production
Practical thinking about architecture, distributed systems, and the decisions that shape reliable software.
Latest writing
Every metric was within threshold. The service was down for half the users. Availability and health are not the same thing.
Transactional outbox is not ceremony — it is how you survive duplicate delivery and partial failures.
What you will find here
Failure modes, recovery paths, and lessons from running systems.
The cost of a decision matters as much as its benefits.
Systems should remain understandable as they grow.