Robert Grossman and Yunhong Gu: Sector: An Open Source Cloud for Data Intensive Computing.

Cloud Computing Conference - Cloud Slam 2009.
Abstract. In this talk, we describe an open source cloud application
called Sector that is designed for data intensive computing. Sector
includes both a compute cloud and a storage cloud. Sector's compute
cloud supports MapReduce, as well as generalizations that allow
user-defined-functions (UDFs) to be invoked easily on all the data
managed by the Sector storage cloud.

In this talk, we give an overview of clouds designed for data
intensive computing, including i) Google's GFS, MapReduce and
BigTable; ii) Hadoop and iii) Sector. We describe some benchmarks for
clouds designed for data intensive computing and include performance
measurements comparing Sector and Hadoop. We also describe some
interoperability issues and emerging standards for these types of
clouds.

Video of the session: