|
Project Information
Featured
Links
|
THIS PROJECT IS PRE-ALPHAIt is not ready for testing yet.We just switched to github and we have decided to use a DHT instead of a meshWhat is ErlFS?ErlFS is a system to store files in a highly redundant and highly available way across multiple machines. ErlFS scales linearly. How does ErlFS work?GeneralErlFS uses distributed Erlang and the chord DHT algorithm. The more nodes you add to the cluster, the more capacity, throughput and redundancy you get. ErlFS is a write-once read-many system. StorageWhen a file is put into the system, it is split up into chunks and each of those chunks are redundantly copied to the next log(N) nodes' file systems. The meta-data includes the keys to each of the chunks for each particular file, content-type and other common attributes. RetrievalWhen a file is retrieved, the client retrieves the meta-data from any node, potentially behind a load-balancer, and uses the list of all the nodes for all of the chunks of a file to retrieve each chunk and joins all of them together to recompose the file, resulting in higher throughput under high load. |