Get the latest tech news

Making a system highly available isn't just about adding redundancy


Written by Shubham Sethi - Senior Software Engineer

To mitigate the cascading impact of such failures, we implemented circuit breakers (CBs) using libraries like hystrix and resilience4j, which are configured to open up and return a default fallback response when error thresholds are breached. Minimal Performance Impact: The introduction of the lightweight Transact Proxy resulted in a negligible increase of ~4 ms in P50 latency at the API Gateway, ensuring system responsiveness remained intact. By addressing key challenges such as data consistency, traffic isolation, dynamic routing, and session stickiness, we’ve built a robust, fault-tolerant system that ensures seamless operations and minimizes user impact during outages.

Get the Android app

Or read this on Hacker News

Read more on:

Photo of System

System

Photo of redundancy

redundancy

Related news:

News photo

Johnny.Decimal – A system to organise your life

News photo

Possible steering failure affects more than 376,000 Teslas from 2023 | The system won't fail while the car is in motion.

News photo

Euro cloud biz trials 'server blades in a cold box' system