
How switching to Variable Bitrate saved 15% bandwidth while improving streaming stability at global scale.
This article explores Netflix's transition from CBR to VBR for all live events, focusing on how they overcame the unpredictability of traffic. It details the adjustments made to traffic steering logic and the fine-tuning of the bitrate ladder using VMAF metrics to ensure consistent quality.
Highly recommended for infrastructure and backend engineers managing large-scale media delivery who want to balance cost efficiency with user experience.
Netflix's live streaming initially relied on Constant Bitrate (CBR), which wasted bandwidth during simple scenes and failed to provide sufficient bits for high-complexity sequences, leading to quality issues.
The team implemented Capped VBR (using AWS QVBR) and redesigned the capacity management system to reserve server resources based on 'Nominal Bitrate' instead of real-time actual throughput to prevent sudden overloads.
Achieved a 15% reduction in average bytes transferred, a 10% drop in peak traffic, and a 5% decrease in rebuffers per hour.
Trade-off
The inherent volatility of VBR makes traffic patterns less predictable, requiring more conservative resource reservations which can slightly limit maximum server utilization.
An encoding method that adjusts the bitrate dynamically based on the complexity of the video content, optimizing for both size and quality.
An open-source perceptual video quality metric developed by Netflix that predicts subjective video quality.
The logic used to route client requests to the most appropriate CDN servers based on available capacity and network conditions.




