The following are the research lines of my current work:
Learning of performance models
Computing systems require the access to peformance information in order to achieve efficien executions of the applications. Machine Learning cames in as a core approach to generate robust performance models with minumim human intervention.
You can access our latest results in this area by making click here.
Workflow Autoscaling and Scheduling
For executing large-scale scientific applications in grids and clouds it is crucial to appropiately managing tasks and resources. Autoscaling permits determining the proper amount and type of cloud instances for executing the applications. Scheduling allows the execution of the applications on the available resources/instances according to a given criteria.