Reproducible and Reusable Data Analysis Workflow Server (Core Infrastructure)
This repository contains the implementation of the core infrastructure for the Reproducible and Reusable Data Analysis Workflow Server (flowServ). This is an experimental prototype to support reuse and evaluation of published data analysis pipelines as well as community benchmarks of data analysis algorithms. flowServ is not yet-another workflow engine. The aim instead is to provide a layer between a client (e.g. a Web user interface) and a workflow engine to facilitate the execution of a defined workflow templates (as shown in the figure below). flowServ is designed to be independent of the underlying workflow engine.
Workflow templates contain placeholders for workflow steps and/or input data and parameters that are provided by the user (e.g., by providing Docker containers that satisfy the workflow steps or uploading input data files). flowServ triggers and monitors the execution of the workflow for the given input values and maintains the workflow results. The API provides the functionality to submit new workflow runs and to retrieve the evaluation results of completed workflow runs.
flowServ was motivated by the Reproducible Open Benchmarks for Data Analysis Platform (ROB). The goal of ROB is to allow user communities to evaluate the performance of their different data analysis algorithms in a controlled competition-style format. In ROB, the benchmark coordinator defines the workflow template along with input data. Benchmark participants provide their own implementation of the variable workfow steps. The workflow engine processes workflows on submission. Execution results are maintained by flowServ in an internal database. The goal of flowServ is to be a more generic platform that can not only be used for benchmarks but also for other types of data analysis workflows.
Workflow templates are motivated by the goal to allow users to run pre-defined data analytics workflows while providing their own input data, parameters, as well as their own code modules. Workflow templates are inspired by, but not limited to, workflow specifications for the Reproducible Research Data Analysis Platform (REANA). The Workflow Templates Section provides further information about templates and their syntax. These templates are used by flowServ to run workflows and to maintain benchmark results.
The flowServ API defines the main interface to programmatically interact with the underlying database and workflow engine. The API implementation that is included in this repository provides a default serialization of all API resources as Python dictionaries. The API is intended to be used by Web applications. These applications can be build using different frameworks. The current default Web API implementation for ROB uses the Flask web framework.
ROB currently provides two different interfaces to interact with a Web API: the Command Line Client and the Web User Interface. See the respective repositories for further information on how to install and use these interfaces.
For an overview of ROB there are slides from the ROB Demo at the Moore-Sloan Data Science Environment’s annual summit 2019 and our presentation at the Analysis Systems Topical Workshop.
The full documentation is also available on readthedocs.io.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size flowserv_core-0.9.1-py3-none-any.whl (260.8 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size flowserv-core-0.9.1.tar.gz (187.8 kB)||File type Source||Python version None||Upload date||Hashes View|
Hashes for flowserv_core-0.9.1-py3-none-any.whl