Notes on How Facebook Handles Its Massive Data Storage

Overview

• Facebook is the world's largest social networking site, valued at over $100 billion.
• It has over a billion users, which is about one in seven people on Earth.
• The company manages a staggering amount of data including profiles, photos, and messages.

Facebook's Origin and Growth

• Founded in 2004 by Harvard students.
• Became a public company and floated on the stock market in 2012 with a valuation of $104 billion.
• Experiences rapid growth of 100 million new users every six months.

Data Management Challenges

• Employee to user ratio: 1 engineer for every 1 million users.
• No existing user guide for such a scale of operation; it is unprecedented.

Data Storage Infrastructure

• Location: Massive data center in Prineville, Oregon, spanning 300,000 square feet.
• Scale: Comparable to a memory chip the size of three football fields.
• Cost: Hundreds of millions of dollars to build.
• Technology: Cutting-edge servers with high-speed data transfer via 21 million feet of fiber optic cables.

Data Retrieval Process

• User requests (e.g., entering Facebook.com) go to the internet and are directed to the data center.
• Information is retrieved from Facebook's servers and sent back to the user in milliseconds.
• Physicality of the Internet: Contrary to the 'cloud' concept, the internet consists of physical buildings interconnected through cables.

Power Supply and Backup

• Data center equipped with 30 megawatts of electricity.
• Backup generators (14 units) can generate 3 megawatts each for power outages.

Cooling Systems

• Importance of cooling to prevent server burnout.
• Uses a seven-room natural air conditioning system:
  • Cool air from Oregon high plains is filtered and mixed to regulate temperature.
  • Mist controls humidity, while large fans expel excess warm air.

User Activity and Growth

• Nearly 600 million users log in daily, nearly double the U.S. population.
• Continual arrival of thousands of new servers to keep up with demand.

Memory Capacity

• Facebook operates with petabytes of data: over 100 petabytes of photos and videos.
• Each server rack can hold 500 terabytes of data.
• Comparatively, this is 130 billion times more memory than the first Apple computer.

Technical Challenges

• Rapid increase in users and data leads to storage becoming a critical issue.
• Technicians like David Gaylord play key roles in maintenance, quickly replacing faulty components.

Conclusion

• With increasing online activity, construction crews are working to expand data center capacity to accommodate future growth.