Date of Award

5-2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Legacy Department

Computer Science

Committee Member

Dr. Amy Apon, Committee Chair

Committee Member

Dr. Brian Malloy

Committee Member

Dr. Jim Martin

Abstract

Amazon Web Services (AWS) provides public cloud computing resources and services and is one of the largest cloud computing providers in the world. However, in order to get started using AWS, one must spend many hours overcoming the steep learning curve and terminology associated with AWS. This is especially true for researchers looking to create and utilize a High Performance Computing (HPC) cluster within AWS. This is due to the massive amount of AWS services and AWS resources that must be created and linked together in order to create a fully functional HPC cluster with AWS. The Dynamic AWS HPC Cluster Project aims to help simplify the steps needed to create a fully functional dynamic HPC cluster within AWS. The user simply completes a simple wizard that specifies the details of the HPC cluster that they want: the size and type of the shared filesystem, the type of HPC scheduler, the number of Compute Instances, what IP addresses they want the cluster to be accessible from, and the number of Login/Head Instances required. After all this has been specified, the Dynamic AWS HPC Cluster project makes the required calls to the AWS APIs in order to create all the required AWS resources. After the resources have been created, they are all automatically configured, networked together, and have the usernames and passwords pushed out to all of the cluster instances for SSH login. The user can then run their jobs and when they have no more jobs left to run they can "pause" the cluster, which means they do not pay for compute charges, and then when they have more jobs to run "resume" the cluster and run their jobs. This allows users to only pay for the cluster when they need it which can help save them money.

Share

COinS