Why is Cloud-Native Application Development Still So Hard?
summary
A prototype distributed OS called Amino run exposes pluggable deployment managers that intercept RPCs to provide replication, caching, offloading and placement without changing application code. It runs on mobile devices and cloud servers, integrates with Kubernetes/etcd, and uses a library of policies. A multiplayer game experiment moves the game-board microservice to a phone to cut read latency for that player, and shows that replicating to both phones speeds reads but raises write latency; a demo and live migration prototype exist.
AI-written summary · about
chapters (24)
- 0:00Introduction
- 0:48Agenda
- 1:42Single User Applications
- 4:48Kubernetes
- 5:24Distributed Systems
- 8:01Application Developer
- 11:10Distributed Cloud Native
- 12:40Outline
- 13:41Goals
- 15:07Amino Run System
- 15:35Amino Run Architecture
- 16:34Application Programming Example
- 17:06Microservices
- 17:58Deployment Kernel
- 19:10Deployment Manager
- 21:19Deployment Managers
- 22:00Deployment Manager Categories
- 22:51Deployment Manager Implementation
- 26:52Microservice
- 27:39Cache
- 29:00Deploy Managers
- 29:26Experiment
- 31:56Languages
- 32:29Next Steps
related talks
30mApp Delivery in Cloud Native: Where are We? What's Next?
35mCloud Native Development Without the Toil
51mCloud Native Development Level Up for Smaller Teams
25mStandardizing Cloud Native Application Delivery Across Different Clouds
41mComplex Application Design
49mApplications architecture spanning multiple clouds
32mCloud Native Progressive Delivery
06mAccelerate Application Delivery with a Cloud-Native Mindset