Boosting System Resilience: Exponential Backoff
summary
Implements exponential backoff with jitter and configurable initial delay, max delay, max attempts and retryable error codes (408, 429, 5xx) against upstreams like MySQL and HTTP services using common clients (Apache HTTP client, Elasticsearch libs). An orchestrator for an Uber Eats–style feed retries a failing data provider up to five times, raising availability to five nines while staying within a 100 ms SLA. Parameters require load-testing to tune for real traffic patterns.
AI-written summary · about
chapters (10)
- 0:00intro
- 0:26understanding exponential backoff
- 1:13real-world example: uber eats
- 2:56handling errors with exponential backoff
- 4:09implementing exponential backoff
- 8:35retryable error codes
- 9:34benefits of exponential backoff
- 10:38libraries supporting exponential backoff
- 11:22personal experience with exponential backoff
- 14:23conclusion and final thoughts
related talks
32mHow does your application behaves when everything goes wrong?
14mHarnessing Cascading Timeouts for System Resilience
34mDistributed Systems - What Can Go Wrong and How to Get It Right
32mSafe Client Behaviour
44mIs Resilience4j enough to make fault tolerant software?
53mResiliency and Availability Design Patterns for the Cloud
38mResilience patterns in practice
21mThe Multi Layered Cake of Resilience