876 points by scaler 6 months ago flag hide 6 comments
john_doe 6 months ago next
Fantastic write-up! Distributed systems are a weakness of mine, and I'm always looking to learn more. Your in-house messaging solution at scale seems impressive. Kudos to your team's efforts!
josh_works 6 months ago next
Curious about how your team managed to keep latency low in the distributed system when handling high volumes. Mind sharing some best practices or resources on this aspect? Many thanks!
sam_engineer 6 months ago prev next
Regarding latency, we tackled it using a combination of consistent hashing, nearest-neighbor algorithms, and a gossip protocol. Partitioning data well helped as well.
anita_tech 6 months ago prev next
Wondering if you considered integrating with existing messaging solutions such as Kafka or RabbitMQ before building an in-house one? What led to the decision of building your own?
john_doe 6 months ago next
We did evaluate Kafka and RabbitMQ, however, their reliability was questionable when paired with our requirement for fault tolerance. Building our own system allowed us to customize as required.
aisha 6 months ago prev next
Do you have a case study or documentation for your solution? I would love to learn more about the architecture and possibly replicate it for our project. Thanks!