Microservices & Kubernetes in a High-demand Environment: The Experience
summary
iFood describes migrating a high-throughput service (processing >20 million records/day) from on-demand instances to container orchestration with Kubernetes, discussing Kubernetes’ pros/cons, limits of AWS Lambda and DynamoStream, and a hybrid solution that reduces ingestion time and lowers Lambda costs by using unused cluster capacity. They also explain scaling from 1 rps to 3,500 rps in under 3 minutes by adopting the Micronaut web framework and simple techniques to achieve sub-1-second application startup, and compare Micronaut to Spring Boot.
AI-written summary · about
chapters (12)
- 0:00Context introduction
- 2:57Growth scaling pains
- 5:04Deploy and config challenges
- 7:58Kubernetes migration
- 11:57Customer service architecture
- 15:59Scalability with HPA
- 17:30Micronaut vs Spring
- 25:02Startup optimization with GraalVM
- 30:10Build and deploy pipeline
- 33:07Monitoring and support challenges
- 35:24Micronaut pros and cons
- 36:10Conclusion and next steps
related talks
18mRe-Architecting Monoliths Into Kubernetes Microservices at Million-User Scale: Hard...
26mThe Journey to Microservice Adoption
59mJava microservices from Netflix OSS to Kubernetes
43mSolving Scalability by Converting a Monolith into Microservices
49mDeveloping microservices applications on the Kubernetes platform
51mMicroservices and Microfrontends: Scaling, Learning, Growing
53mNext Generation Microservices
40mMicronaut in Production