Netflix’s Journey: From DVD Rental to Cloud‑Native Chaos Engineering on AWS

Previously I mentioned Netflix only in the context of "House of Cards"; while studying DevOps Professional I discovered that Chaos Monkey originated from Netflix, along with Eureka and Hystrix. Curious, I researched and compiled recent Netflix‑related material into this article.

Most of the content is sourced from the references listed at the end, with adjustments and personal commentary.

What is Netflix Netflix is a video‑streaming company offering subscription plans ranging from $7.99 to $11.99 per month. It operates an "all‑you‑can‑eat" model and has about 81 million subscribers worldwide, over 46 million of whom are in the United States.

The Origin of Netflix Long ago Blockbuster dominated the video‑rental market. Reed Hastings, after being hit with a large overdue fee, was inspired by the gym membership model and, being a software engineer with capital, founded Netflix as a DVD‑rental service without late fees. Thirteen years later Netflix drove Blockbuster into bankruptcy.

Netflix transformed Blockbuster’s model by (1) adopting a light‑asset, online‑only approach and (2) mailing DVDs to customers, essentially creating a hybrid e‑commerce/O2O service.

2006 marked Netflix’s streaming era. With broadband expansion and YouTube’s rise, Netflix shifted from physical rentals to online content delivery, growing from 4.2 million subscribers in 2005 to 83.2 million in 2016 – a ten‑fold increase in a decade, illustrating the cost‑advantage and scale of a cloud‑based service.

Beyond its entertainment brand, Netflix is a technology company. It pioneered the open‑source community with projects like Eureka, Hystrix, and the Simian Army, and maintains a dedicated Netflix Tech Blog discussing cutting‑edge engineering challenges.

Netflix’s Chaos Monkey Army To ensure resilience, Netflix runs Chaos Monkey on AWS, randomly terminating services during work hours to test production stability. This concept evolved into the Simian Army, which includes:

Chaos Monkey : randomly kills production instances.

Latency Monkey : injects artificial delays into REST calls.

Conformity Monkey : shuts down instances that violate best‑practice configurations.

Doctor Monkey : removes unhealthy instances.

Janitor Monkey : reclaims unused resources.

Security Monkey : scans for security vulnerabilities and validates certificates.

10‑18 Monkey : checks localization and internationalization settings.

Chaos Gorilla : simulates an entire AWS Availability Zone failure.

Netflix also contributes many open‑source projects:

Common Runtime Services & Libraries (e.g., Eureka, Ribbon, Hystrix)

Big Data tools (e.g., Genie)

Build and Delivery tools (e.g., Asgard, Spinnaker)

Data Persistence (e.g., EVCache)

Reliability & Performance (e.g., Simian Army)

These projects, especially Eureka, Hystrix, and the Simian Army, are widely recognized.

Netflix and AWS Netflix is one of AWS’s most important customers, accounting for over one‑third of North American internet traffic that traverses AWS. Its reliance spans compute, storage, big‑data analytics, and AI. Netflix frequently presents at AWS re:Invent; in 2015, eight Netflix engineers spoke.

In 2008 a database corruption caused a three‑day outage, prompting Netflix to migrate its entire infrastructure to AWS for reliability. The move required custom tooling; Netflix built an auto‑deployment system (now with >2100 GitHub stars) long before AWS released CodeDeploy.

Key lessons from Netflix’s AWS journey:

What works in a private data center may not translate directly to the cloud.

Cloud design must account for multi‑tenant resource sharing and variable throughput.

Netflix’s “Lambo Architecture” aims for every service to survive total system loss.

Real‑scale testing in production environments is essential.

Conclusion: Antifragility Embracing chaos, volatility, and risk—through tools like Chaos Monkey—makes systems stronger. Regularly inducing failures, automating responses, and learning from small injuries embody the antifragile principle.

References:

https://www.zhihu.com/question/19552101/answer/114867581

https://zhuanlan.zhihu.com/p/19681894

https://my.oschina.net/moooofly/blog/828545