Previous webinars have shown how to automate the orchestration of an experiment from allocating the resources to building the lab to running the experiment and tearing it all down using Heat Templates.
Chameleon is a research platform in which experimenters can make on-demand or advanced reservations for resources. Some experiments require resources — GPUs, FPGAs — that are both scarce and popular so users must reserve them well in advance. Other experiments require large portions of the testbed and are ideally scheduled during off-peak times. In all cases, efficient use of allocated resources requires experiments to start as soon as the reservation becomes active. It is often difficult for users to be physically available to start an experiment for reservations made far in the future or during off-peak hours.
This webinar will demonstrate how to use heat templates to automatically deploy experiments when a reservation becomes active. We will use Chameleon’s new Automatic Deployment mechanism to start a reservation and deploy the experiment without human intervention.
You should have some experience with Heat Templates (https://www.chameleoncloud.org/news/orchestration-heat-templates/) before attending this webinar.
© 2024 RENCI
