Dynamo Explained: How Amazon Built a Highly Available Distributed System скачать в хорошем качестве

Dynamo Explained: How Amazon Built a Highly Available Distributed System

SEO Description Introduction: In this video, we explore the design and implementation of Dynamo, a highly available and scalable distributed data store built to power Amazon’s e-commerce platform. At Amazon's scale, components fail continuously; Dynamo was engineered to treat these failures as the "normal case" without impacting the customer experience. Key Concepts Covered: • The "Always-On" Requirement: Learn how Dynamo provides a simple primary-key-only interface to ensure customers can always view and add items to their shopping carts, even during disk failures or network outages. • Sacrificing Consistency: Discover why Dynamo prioritises high availability over strict consistency, using eventual consistency to maintain performance under stress. • Technical Architecture: We break down the synthesis of techniques used in the system, including consistent hashing for incremental scalability, vector clocks for object versioning, and gossip-based protocols for failure detection. • Service Level Agreements (SLAs): Understand Amazon’s relentless focus on performance at the 99.9th percentile, ensuring that even the most demanding customer requests are handled within strict latency bounds. Why It Matters: The Dynamo paper is a foundational text in distributed systems. It demonstrates how decentralized techniques can be combined to provide a single, highly available system that handles extreme peak loads during the busy holiday shopping season without downtime Paper Link - https://assets.amazon.science/ac/1d/e...