SOA Synthetic Reliability Models


Datasets contain reliability models for composite service-oriented applications. The models were built using 30 atomic services, while taking into account sequence, parallel, branch and loop workflow component. Each dataset contains 5000 models.



Atomic services

  • 30 atomic services were generated
  • Reliability for each component was genereted using a normal distribution with the mean value 0.99 and standard deviation 0.01

Build type

  • 30-build-all: Each model was constructed by using all of the 30 components
  • 30-build-10:Each model was constructed by using 10 randomly selected components (uniform distribution)
  • 30-build-5-10: Each model was used by choosing 5-10 randomly selected components (uniform distribution)

WF Components

  • all: All workflow components were user with following probabilities (sequence 0.7, parallel 0.1, loop 0.1, branch 0.1)
  • sequence only: Only the sequence components are used to construct the model


  • high: Each model is constructed to have 5-10 levels, each level having 6-12 nodes (uniform distribution)
  • low: Each model is constructed to have 2-5 levels, each level having 3-6 nodes (uniform distribution)
Build Type WF Components Complexity Download link
30-build-all all high Download
30-build-all all low Download
30-build-all sequence only high Download
30-build-all sequence only low Download
30-build-10 all high Download
30-build-10 all low Download
30-build-10 sequence only high Download
30-build-10 sequence only low Download
30-build-5-10 all high Download
30-build-5-10 all low Download
30-build-5-10 sequence only high Download
30-build-5-10 sequence only low Download

Generating Additional Datasets

Datasets can be generated by running the model generator available at this link. You can find the source code for the generator here.

It is necessary to install java jre in order to run the generator. The is run using the following command:

java -jar modelGenerator.jar <generateImages>

The generateImages parameter can be used to define if generator is to create model imagaes or not (value is true or false).

Configuration file for the generator named dataset.json has to be in the same folder as modelGenerator.jar. The output is generated in the folder other. An example of the dataset.json file is available here.

Citing RSOA Datasets

We would kindly ask you to cite our datasets if you found them useful in your work. You can use the following citation format (BIBTex):

  author       = {Goran Delac, Marin Silic, Klemo Vladimir and Ivan Budiselic},
  title        = {{RSOA Datasets}: Synthetic Reliability Models},
  howpublished = {\url{}},
  month        = January,
  year         = 2017

Comments are closed.