Prediction

Most datacenters, clouds and grids consist of multiple generations of computing systems, each with different performance profiles, posing a challenge to job schedulers in achieving the best usage of the infrastructure. A useful piece of information for scheduling jobs, typically not available, is the extent to which applications will use available resources once they are executed. This project investigates suitability of several machine learning techniques for predicting spatiotemporal utilization of resources by applications.

Data collected from real experiments executing bioinformatics applications BLAST and RAxML:


 * RAxML dataset for predicting RSS
 * RAxML dataset for predicting Time
 * BLAST dataset for predicting Output
 * BLAST dataset for predicting Time