As part of an effort to expand access to computing resources, the (US) National Science Foundation XSEDE project has funded several new computational resources with cloud provisioning capabilities – Bridges and JetStream. To facilitate sharing and reuse of scientific applications, an XSEDE Cloud Virtual Machine Repository has been proposed. Traditional delivery of applications involved monolithic images that requires development, construction, testing, maintenance, vetting, and cataloging of cloud virtual machine images which are nontrivial tasks. One solution is to use Murano, Heat, and cloud-init scripts to deliver scientific applications on a small number of generic images from Linux distribution maintainers rather than images. The use of standard OpenStack components makes the application repository usable by any standard OpenStack deployment. The details of construction, use, and highlights of a few selected applications will be covered.
Attendees should expect to learn the discovered best practices for creation and maintenance of a scientific application repository using Murano, Heat templates, and cloud-init user data scripts. They should also learn how to use and contribute to the (US) National Science Foundation’s XSEDE project cloud application repository.