Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Workflow Generator

To facilitate evaluation of workflow algorithms and systems on a range of workflow sizes, we have developed a set of synthetic workflow generators. These generators use the information gathered from actual executions of scientific workflows on the Grid as well as our understanding of the processes behind these workflows to generate realistic, synthetic workflows resembling those used by real world scientific applications.

Code

The code used to generate all of the synthetic workflows below, and many others, is available from the GitHub repository. The java workflow generator sometimes generates negative task runtimes, so watch out for that.

Simulator

WorkflowSim can be used to simulate the workflows generated by the WorkflowGenerator.Workflow Generator.

Traces

Traces and execution logs from real workflows are available here: herehere, and here. Data sets like these were used to parameterize the Workflow Generator.

Synthetic Workflows

Pegasus Workflows

These workflows come from a paper by Bharathi, et al. [1]. There is another paper with more information about the workflows by Juve, et al. [2].

...

Workflow Type

Example

DAX

Montage
The Montage application created
by NASA/IPAC stitches together multiple
input images to create
custom mosaics of the sky.

25 Node DAX
50 Node DAX
100 Node DAX
1000 Node DAX

CyberShake
The CyberShake workflow is used
by the Southern Calfornia Earthquake
Center to characterize
earthquake hazards in a region.

30 Node DAX
50 Node DAX
100 Node DAX
1000 Node DAX

Epigenomics
The epigenomics workflow created
by the USC Epigenome Center
and the Pegasus Team is used to
automate various operations
in genome sequence processing.

24 Node DAX
46 Node DAX
100 Node DAX
997 Node DAX

LIGO Inspiral Analysis
LIGO's Inspiral Analysis workflow
is used to generate and
analyze gravitational waveforms
from data collected during the
coalescing of compact binary systems.

30 Node DAX
50 Node DAX
100 Node DAX
1000 Node DAX

SIPHT
The SIPHT workflow, from the
bioinformatics project at Harvard,
is used to automate the search for
untranslated RNAs (sRNAs) for bacterial
replicons in the NCBI database.

30 Node DAX
60 Node DAX
100 Node DAX
1000 Node DAX

Ramakrishnan and Gannon Workflows

These workflows come from a report by Ramakrishnan and Gannon [3].

...