Load Balancing and Consistent Hashing
My demonstration lecture is about load balancing via consistent hashing. Consistent hashing is a technique, originally developed for use in the Akamai content delivery network, for dividing data among multiple servers such that clients know which server is responsible for a given piece of data without communication and despite servers being added and removed from the system. Consistent hashing, or a variant thereof, is used in Amazon's Dynamo database, the HAPRoxy load balancer, and many distributed hash tables. My will explain the consistent hashing algorithm in detail and discuss the challenges of load balancing with real data, where some keys will be accessed many orders of magnitude more than others.
Doug Woos is a PhD candidate in Computer Science at the University of Washington, where he is advised by Tom Anderson, Mike Ernst, and Zach Tatlock. His current research focuses on helping developers build reliable distributed systems. His previous work spans the areas of systems, networks, and programming languages. Before starting at UW, he obtained a B.A. from Swarthmore College.
Host: Professor Kathi Fisler