Explore Netflix's journey in building a globally distributed cloud application in this 58-minute conference talk from GOTO Aarhus 2012. Dive into the company's rapid growth and migration to AWS, the integration of Apache Cassandra, and the internationalization of their streaming service. Learn about Netflix's custom PaaS built on AWS, utilizing Java/Tomcat, Groovy/Grails, and Python. Discover key architectural patterns for availability, including isolated services, balanced availability zones, and triple replicated persistence. Examine the transition from datacenter to cloud, the implementation of Chaos Monkey, and the use of circuit breakers for resilient design. Gain insights into Netflix's build pipeline, Jenkins architecture, and open-source projects. Understand cloud access control, runtime considerations, and future roadmap plans. This comprehensive talk covers everything from high-level architecture to specific tools and practices, providing valuable knowledge for building scalable, resilient cloud applications.
Globally Distributed Cloud Application at Netflix
Overview
Syllabus
Introduction
What Netflix Did
Keeping up with Developer Trends
Portability vs. Functionality
Freedom and Responsibility
Member Web Site
Content Delivery Service
Streaming Device API
Major Front End Services
Component Services
Web Server Dependencies Flow
One Request Snapshot
Current Architectural Patterns for Availability
Isolated Services
Three Balanced Availability Zones
Triple Replicated Persistence
Isolated Regions
Failure Modes and Effects
Netflix Deployed on AWS
Datacenter to Cloud Transition Goals
Netflix Datacenter vs. Cloud Arch
Chaos Monkey
Responsibility and Experience
Resilient Design - Circuit Breakers
Components
Build Pipeline
Jenkins Architecture
Netflix Extensions to Jenkins
The Bakery
Accounts Isolate Concerns
Reservations and Billing
Cloud Access Control
Runtime
Open Source Projects
Roadmap for 2012
Amazon Cloud Terminology Reference
Taught by
GOTO Conferences
