Using Machine Learning to Optimize a Workflow Management System.

Workflow management frameworks support the creation of task dependencies and make efficient use of resources while running those workloads. Typically, these tasks can be long running processes like machine learning algorithms or access data from databases. Workflow management consists of mapping tasks to suitable resources and the management of workflow execution in a cloud environment. The goal of this project is to optimize the job scheduling algorithm using machine learning techniques in a workflow orchestration framework that manage workloads across a heterogeneous system. Our proposed approach applies a machine learning algorithm to workflow event logs to learn properties of resources required to perform a task. When a new process is initiated, the trained classifier can suggest a suitable resource to undertake the specified task. Adding these features to Rubikloudâs machine learning pipeline would improve the efficiency and scalability of the existing machine learning infrastructure. TO BE CONT’D

Faculty Supervisor:

Eyal de Lara

Student:

Nisal Perera

Partner:

Rubikloud Technologies Inc.

Discipline:

Computer science

Sector:

Information and communications technologies

University:

Program: