Please join our guest speaker Michael Isard visiting from the Microsoft Research Silicon Valley Lab. He will be speaking on Monday, 2/4, 1-2PM, Engineering 2, room 215.

Dryad: Distributed Data-Parallel Programs from Sequential Building Blocks

Dryad is a general-purpose distributed execution engine for coarse- grain data-parallel applications. A Dryad application combines computational “vertices” with communication “channels” to form a data- flow graph. Dryad runs the application by executing the vertices of this graph on a set of available computers, communicating as appropriate through files, TCP pipes, and shared-memory FIFOs. The Dryad C++ APIs support very general computations and dynamic optimization strategies. This generality brings some complexity, however Dryad is designed to be suitable as a “middleware” target for simpler, higher-level programming abstractions. Several higher-level layers have already been developed, and I will briefly outline DryadLINQ which offers a declarative data-parallel programming model, integrated with C#.

Michael Isard has been a researcher at Microsoft Research Silicon Valley since 2002. His research background was originally in computer vision and statistical inference, but most of his current work is in systems. Several recent projects focus on programming models that make it easier to exploit parallel and distributed computing resources.

Host: Carlos Maltzahn