Cluster I/O with River

Abstract

imum performance in the c&on case -even in the-race a* nor,uniformities in hardware, software, and workload. River is based on two simple design features: a high-performanccdirniburedqueue, and a storage redundancy mechanism called gradunfed declusrering. We have implemented a number of data-intensive applications on River, which validate our design with near-ideal pxformancc in B variety of n&uniform performance scenarios.