Member-only story
System Design of Netflix
Netflix is a global entertainment giant with over 200 million subscribers. To comprehend the architecture of a system supporting such a large user base, we’ll start with some back-of-the-envelope calculations.
Assume at peak times, about 25% of subscribers, approximately 50 million users, are actively streaming. If we consider each stream to consume about 3GB for a 2-hour movie, and this data is evenly distributed over the movie’s duration, we calculate:
- Data streamed per second: (50 million streams * 3GB/stream) / (2 hours/stream * 3600 seconds/hour) ≈ 10.42 GB/s
Due to Netflix’s worldwide presence, data must be replicated across various geographical locations to ensure redundancy and low latency. If we assume an average replication factor of 3, the overall data output requirement increases to about 31.25 GB/s.
These calculations underline the massive scale of Netflix’s operations, providing context for the subsequent system design discussions.
Open Connect (OC) — Netflix’s Content Delivery Network (CDN)
To efficiently handle this huge volume of data, Netflix uses its own Content Delivery Network (CDN) known as Open Connect (OC). The OC includes original servers and edge servers. Edge servers are strategically placed at Internet Service Providers (ISP) to enable swift data delivery.
Question: What advantages does Netflix gain by hosting edge servers at ISPs?
Answer: Hosting…
