Chaos Engineering
Chaos engineering deliberately injects failures — killing servers, adding network latency, exhausting resources — into a system, often in production, to proactively discover weaknesses before they cause a real, unplanned outage. It treats resilience as something to continuously verify rather than assume.
Popularized by Netflix's "Chaos Monkey" (which randomly terminates production instances to force teams to build genuinely fault-tolerant systems), the practice inverts the usual assumption that failure testing happens only in a controlled test environment — real production complexity is often impossible to fully replicate elsewhere, so testing resilience there directly, carefully and incrementally, surfaces issues staging environments never would.
Mature chaos engineering practice runs controlled experiments with a clear hypothesis ("the system should route around a failed availability zone within 30 seconds"), a defined blast radius to limit real user impact, and an immediate abort mechanism — it's a deliberate, disciplined practice, not just randomly breaking things and hoping for the best.
