Netflix at AWS re:Invent 2019

Netflix Technology Blog
7 min readintermediate
--
View Original

Overview

The article discusses Netflix's participation in AWS re:Invent 2019, highlighting their speaking events and sessions focused on technology advancements and operational strategies. Key topics include Amazon EC2 Auto Scaling, the Netflix engineering culture, and innovative solutions for managing data and application deployment at scale.

What You'll Learn

1

How to utilize Amazon EC2 Auto Scaling for capacity management

2

Why Netflix's engineering culture enhances operational efficiency

3

How to implement a file system as a service using Ceph

4

How to benchmark stateful services in AWS effectively

Prerequisites & Requirements

  • Understanding of cloud computing concepts
  • Familiarity with AWS services(optional)
  • Experience with performance engineering(optional)

Key Questions Answered

How does Netflix manage capacity with Amazon EC2 Auto Scaling?
Netflix utilizes Amazon EC2 Auto Scaling to maintain a healthy fleet, improve application availability, and reduce costs. This hands-free capacity management allows Netflix's infrastructure to automatically adapt to changing traffic patterns, ensuring a seamless viewing experience for over 150 million subscribers globally.
What technologies does Netflix use to support its engineering culture?
Netflix employs a variety of technologies to support its engineering culture, including open-source tools like Spinnaker for continuous delivery and Ceph for file storage. This approach enables engineers to work autonomously and efficiently, fostering innovation and rapid deployment of applications.
What are the challenges of benchmarking stateful services in AWS?
Benchmarking stateful services in AWS presents unique challenges due to strict sub-millisecond tail-latency requirements. Netflix shares its philosophy and lessons learned from operating these services, emphasizing the importance of benchmarking in maintaining performance across multiple regions.
How does Netflix monitor anomalous application behavior?
Netflix monitors anomalous application behavior using AWS CloudTrail to track activities across multi-account environments. This system alerts teams to unusual actions, enabling proactive security measures and operational oversight.

Key Statistics & Figures

Number of Netflix subscribers
150 million
Netflix serves this number of subscribers across 190+ countries, highlighting the scale at which their infrastructure must operate.

Technologies & Tools

Cloud Computing
Amazon EC2
Used for auto-scaling and managing capacity.
Storage
Ceph
Deployed for file system as a service to manage increasing data needs.
Deployment
Spinnaker
An open-source continuous delivery platform created by Netflix.
Monitoring
AWS Cloudtrail
Used for tracking application behavior and monitoring anomalous actions.
Performance Analysis
Extended Bpf (ebpf)
Utilized for creating performance-analysis tools in the cloud.

Key Actionable Insights

1
Implement Amazon EC2 Auto Scaling to dynamically adjust your infrastructure based on traffic patterns.
This approach not only optimizes resource usage but also ensures high availability for applications, which is crucial for maintaining user satisfaction during peak traffic times.
2
Leverage open-source tools like Spinnaker for continuous delivery to enhance deployment efficiency.
By adopting such tools, teams can reduce the time to market for new features, allowing for quicker responses to user feedback and market changes.
3
Utilize benchmarking as a critical part of your system's lifecycle to maintain performance standards.
Regular benchmarking helps identify performance bottlenecks early, ensuring that services meet the required latency and throughput as they scale.

Common Pitfalls

1
Neglecting the importance of capacity management can lead to application downtime during peak traffic.
Without proper scaling mechanisms like Amazon EC2 Auto Scaling, applications may struggle to handle sudden increases in user demand, resulting in poor user experiences.

Related Concepts

Cloud Computing Strategies
Continuous Delivery Practices
Performance Engineering Techniques