How To Build a Distributed System (And Should You?)

on YouTube ↗38m

" summary

A geo-distributed global directory service is built and deployed to clusters in the US, Germany, and Singapore using gRPC, Kubernetes, LevelDB and Prometheus; two open-source components honu and turtle provide metadata versioning and bilateral anti-entropy replication, delivering eventual consistency so replicas converge without Spanner/Aurora costs. Engineers hit concrete failures: protocol buffer decoding errors, tangled send/receive goroutines, and incorrect replication phase handling, but experiments tuning anti-entropy intervals reduce staleness and improve recovery.

AI-written summary · about

" chapters (7)

1. 0:00Building distributed systems
2. 2:33Global directory service story
3. 6:50What defines a distributed system
4. 11:46Kubernetes deployment lessons
5. 17:51Geoping DNS routing demo
6. 22:25Future of distributed systems
7. 28:07Q&A

" related talks

• 37mHow Geo-Distributed Apps Use Global Cloud InfraDenis Magda · Conf42 Cloud Native 2022
• 30mDistributed system designBdelaziz Eroui · BlaBlaConf 2022
• 11mExplaining distributed systems like I'm fiveMichele Riva · LeadDev Berlin 2022
• 46mBuilding high available systems with Akka Distributed DataJeroen Gordijn · Scala Days Copenhagen 2017
• 01h 00mThe Secret Life of Distributed SystemsOleksii Petrov · Software Architecture fwdays'25
• 20mScaling Microservices and Distributed Databases in a Multi-Cloud WorldRam Ravichandran · Distributed SQL Summit 2019
• 59mHow to Escape The Distributed MonolithIan Cooper · NDC Minnesota 2019
• 48mOnline OLTP Computing and Traffic Governance as a Service for True Digital TransformatioTrista Pan · DevDays 2023