SwarmForm 1.0.2

SwarmForm

SwarmForm is a distributed workflow management system for High Performance Computing environments. SwarmForm introduces task clustering to reduce the makespan of workflows executed using it by minimizing the overheads. SwarmForm follows a distributed architecture with independent programs controlling different functional layers in workflow management.

Getting Started

These instructions will get you familiar with the installation and basic functions of the SwarmForm.

_Prerequisites_

Following prerequisites should be installed to run SwarmForm.

• MongoDB

• Python 3.3+

• pip

_Installation_

To install SwarmForm, simply type

` pip install SwarmForm `

_Basic Usage_

Following is a step by step series of examples that tell you how to use the basic system functions.

• Initialize a SwarmForm launchpad YAML file

` sform init `

• Insert a SwarmFlow from file

` sform add -sf <file path> `

• Get SwarmFlow from SwarmPad

` sform get_sf -id <SwarmFlow ID> `

• Cluster the fireworks in the SwarmFlow and save the new SwarmFlow to the database

` sform cluster -sf <SwarmFlow ID> ` or

` sform cluster -sf <SwarmFlow ID> -a <clustering algorithm rac/wpa> -cc <cluster count> `

• Reset and re-initialize the SwarmForm database

` sform reset `

Built With

• [Python](https://www.python.org/)

• [FireWorks](https://github.com/materialsproject/fireworks)

Authors

• Kalana Dananjaya - https://github.com/KalanaDananjaya

• Ayesh Weerasinghe - https://github.com/AyeshW

• Randika Jayasekara - https://github.com/rpjayasekara

Publication

A. Weerasinghe, K. Wijethunga, R. Jayasekara, I. Perera and A. Wickramarachchi, “SwarmForm: A Distributed Workflow Management System with Task Clustering”, in 20th International Conference on Advances in ICT for Emerging Regions, Colombo, Sri Lanka, 2020.

Copyright 2021 © SwarmForm.