What I’ll write about

I work on systems that have to stay up — distributed databases, message queues, deployment pipelines, the operational glue around all of it. Posts here will mostly cover:

• Failure analysis — postmortems abstracted of names and blame.
• Reliability patterns — backpressure, idempotency, exactly-once illusions, and where their assumptions break.
• Tooling — small notes on instruments I find indispensable.

A code sample, for testing

def with_retry(fn, attempts=3, backoff=0.2):
    for i in range(attempts):
        try:
            return fn()
        except TransientError:
            if i == attempts - 1:
                raise
            time.sleep(backoff * (2 ** i))

A short inline code snippet for good measure.

A blockquote

Anything that can go wrong, will go wrong — at exactly the moment your on-call rotation hands off.

That’s all for now.