Amazon's Dynamo
Amazon's CTO introduces Dynamo, a highly available and scalable distributed data store built for Amazon's
platform in a paper that he has co-authored for SOSP. Dynamo is not S3 in case you're wondering. It's an internal data store, not available to the public via web services.
The significance of the paper is the degree of high availability required for Dynamo, and how that requirement is addressed in Dynamo's architecture design. Companies like Yahoo!, Google and Amazon have to deal with failure constantly. The number of moving parts in their infrastructure is so vast that at any given time you can see disks, servers, or network devices failing. Technologies like Dynamo are built to make such failures transparent to the end uses of the system.
A PDF version of the paper is also available. It's a long paper, but a great read for anybody interested in HA and distributed systems - this is not academic stuff. It's a production system, which is rare and therefore very valuable.
The significance of the paper is the degree of high availability required for Dynamo, and how that requirement is addressed in Dynamo's architecture design. Companies like Yahoo!, Google and Amazon have to deal with failure constantly. The number of moving parts in their infrastructure is so vast that at any given time you can see disks, servers, or network devices failing. Technologies like Dynamo are built to make such failures transparent to the end uses of the system.
A PDF version of the paper is also available. It's a long paper, but a great read for anybody interested in HA and distributed systems - this is not academic stuff. It's a production system, which is rare and therefore very valuable.
Categories
Technical0 TrackBacks
Listed below are links to blogs that reference this entry: Amazon's Dynamo.
TrackBack URL for this entry: http://www.unixica.com/mt4/mt-tb.cgi/70
Leave a comment