Migrating to Cloud Native with Microservices
summary
Moves encoding and logging workloads to AWS first, then shifts the entire customer-facing site to cloud by Christmas 2010; uses S3 + EMR for logs and Hadoop, runs 3–4k machines for encoding, and proves heavy I/O with a one-off 96 nodes with 2TB SSD each that ingests 18TB and is torn down days later. Replaces relational DBs with about 10–15 Cassandra clusters (per service), runs hundreds of independent services (examples show ~450), uses chaos testing and containers (Docker) and one-second metrics to validate that the approach works.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 0:30Netflix
- 2:10AWS
- 3:35Cassandra
- 5:00AWS vs Netflix
- 6:10Leaving Netflix
- 6:45Optimize for speed
- 7:15Culture
- 7:45Capacity
- 9:25Architecture
- 11:15Clusters
- 14:00Enterprise adoption curve
- 15:10Enterprise finally embracing cloud
- 16:25DevOps Enterprise Summit
- 17:35What separates incumbents from disruptors
- 18:25Processes prevent problems
- 19:50The Goal
- 21:30Product Development Process
- 24:30Silos
- 27:25Monolithic App
- 30:32Product Updates
- 34:15Why Microservices
- 34:40Definition of Microservices
- 42:30Monitoring Microservices
- 47:27Wrap up
related talks
28mCloud The Final Destination
22mFireside Chat with AWS
48mThe Journey of a strongly coupled Monolith into the Cloud Land
30mLarge-Scale Media Archive Migration to the Cloud
25mAutobahn in the cloud
20mScaling Microservices and Distributed Databases in a Multi-Cloud World
33mNavigating the Seas of Transformation: Migrating from EC2 to Kubernetes
20mMigration of 300 microservices from AWS Cloud to