> Term
Retry Storm
A cascading failure caused when clients aggressively retry failed requests, overwhelming a recovering service.
Detailed Explanation
When a service blips, clients naturally retry. If millions of clients retry at the exact same millisecond, they create a self-inflicted DDoS attack. The recovering service is instantly crushed by the backlog of retries combined with new traffic, preventing it from ever stabilizing.
This is why exponential backoff and jitter are mandatory in distributed systems.
Why It Matters
It turns a 5-second network hiccup into a 2-hour catastrophic total system failure.
Common Failure Mode
Practical Example
Production Manifestation
A database restarting, immediately hitting 100% CPU, and crashing again in an infinite loop.
Frequently Asked Questions
What is Retry Storm in short?
A cascading failure caused when clients aggressively retry failed requests, overwhelming a recovering service.
What is the most common failure mode?
Hardcoding a simple `while(failed) { retry(); }` loop in a mobile client distributed to a million users.
AI Summary
A cascading failure caused when clients aggressively retry failed requests, overwhelming a recovering service. It turns a 5-second network hiccup into a 2-hour catastrophic total system failure.