Dryad was a research project atMicrosoft Research for a general purpose runtime for execution ofdata parallel applications. The research prototypes of the Dryad and DryadLINQ data-parallel processing frameworks are available in source form atGitHub.^[1]

Microsoft made several preview releases of this technology available as add-ons toWindows HPC Server 2008 R2.

An application written for Dryad is modeled as adirected acyclic graph (DAG). The DAG defines thedataflow of the application, and the vertices of the graph defines the operations that are to be performed on the data. The "computational vertices" are written using sequential constructs, devoid of anyconcurrency ormutual exclusion semantics. The Dryad runtime parallelizes the dataflow graph by distributing the computational vertices across various execution engines (which can be multiple processor cores on the same computer or different physical computers connected by a network, as in acluster). Scheduling of the computational vertices on the available hardware is handled by the Dryad runtime, without any explicit intervention by the developer of the application or administrator of the network. The flow of data between one computational vertex to another is implemented by using communication "channels" between the vertices, which in physical implementation is realized byTCP/IP streams,shared memory or temporaryfiles. A stream is used at runtime to transport a finite number ofstructuredItems.

Dryad defines adomain-specific language, which is implemented via aC++ library, that is used to create and model a Dryad execution graph. Computational vertices are written using standard C++ constructs. To make them accessible to the Dryad runtime, they must be encapsulated in aclass thatinherits from theGraphNode base class. The graph is defined by adding edges; edges are added by using a composition operator (defined by Dryad) that connects two graphs (or two nodes of a graph) with an edge.Managed code wrappers for the DryadAPI can also be written.

There exist several high-level language compilers which use Dryad as a runtime; examples includeScope (Structured Computations Optimized for Parallel Execution) and DryadLINQ.^[2]

In October 2011, Microsoft discontinued active development on Dryad, shifting focus to theApache Hadoop framework.^[3][4][5]

• "Dryad: Distributed Data-Parallel Programs from Sequential Building Blocks"(PDF).Microsoft Research. Retrieved2007-12-04.
• "SCOPE: Easy and Efficient Parallel Processing of Massive Data Sets"(PDF).Microsoft Research. Retrieved2009-01-21.

• Dryad: Programming the Data CenterArchived 2007-09-12 at theWayback Machine
• Dryad Home
• Video of Michael Isard explaining Dryad at Google

• Concurrent programming libraries
• Distributed computing architecture
• Microsoft free software
• Microsoft Research
• Software using the Apache license
• Windows-only free software

• Articles with short description
• Short description matches Wikidata
• Webarchive template wayback links