Dryad
E702187
Dryad is a general-purpose distributed execution engine from Microsoft Research designed to efficiently run data-parallel applications across large clusters of computers.
Statements (47)
| Predicate | Object |
|---|---|
| instanceOf |
data-parallel execution engine
ⓘ
distributed execution engine ⓘ |
| affiliation | Microsoft Research NERFINISHED ⓘ |
| communicationModel | explicit data channels between vertices ⓘ |
| componentOf | Microsoft Research distributed systems projects ⓘ |
| computingParadigm |
data-parallel computing
ⓘ
distributed computing ⓘ |
| dataModel | immutable data passed between vertices ⓘ |
| designedFor |
general-purpose distributed execution
ⓘ
large clusters of computers ⓘ |
| developer | Microsoft Research NERFINISHED ⓘ |
| distinctionFromMapReduce | uses general DAGs instead of a fixed map-reduce pipeline ⓘ |
| executionModel |
directed acyclic graph of computational vertices
ⓘ
graph-based execution model ⓘ |
| faultToleranceMechanism | restarting failed vertices ⓘ |
| feature |
automatic scheduling of vertices across machines
ⓘ
dynamic graph optimization at runtime ⓘ fault tolerance through re-execution of vertices ⓘ separation of application logic from execution policy ⓘ |
| granularity | coarse-grain tasks ⓘ |
| influenced | DryadLINQ NERFINISHED ⓘ |
| introducedIn | mid-2000s ⓘ |
| languageSupport |
C#
NERFINISHED
ⓘ
C++ ⓘ |
| license | proprietary ⓘ |
| notablePublication | Dryad: Distributed Data-Parallel Programs from Sequential Building Blocks NERFINISHED ⓘ |
| optimization | data locality-aware scheduling ⓘ |
| organization | Microsoft NERFINISHED ⓘ |
| purpose |
execute applications on large clusters
ⓘ
run data-parallel applications ⓘ |
| relatedWork |
DryadLINQ
NERFINISHED
ⓘ
MapReduce NERFINISHED ⓘ |
| researchArea |
cluster computing
ⓘ
distributed systems ⓘ parallel processing ⓘ |
| runsOn |
Windows-based clusters
ⓘ
cluster of commodity computers ⓘ |
| scheduling | centralized job manager ⓘ |
| supports |
batch processing workloads
ⓘ
coarse-grain data-parallel applications ⓘ large-scale data analysis ⓘ |
| type |
cluster computing framework
ⓘ
dataflow execution engine ⓘ |
| uses |
TCP pipes
ⓘ
shared-memory channels ⓘ temporary files as communication channels ⓘ vertices and channels to form a dataflow graph ⓘ |
Referenced by (2)
Full triples — surface form annotated when it differs from this entity's canonical label.