Virtual Machine and Data Placement Based on Physical Particle Optimization Model for Running Remote Sensing Scientific Flows

Nowadays scientific geographical image processing data is growing more than ever thanks to improvements in satellite sensor resolution. The major challenge is to consume large amounts of sensor data. We propose a system that is based on the fact that both geographic data and the computation are intelligently placed in cloud data centers. Our solution is scheduling virtual machines and data sources across a large cloud data center. The VM and data placement solution is simulated by simple physical particles that interact in various ways and at various times. Particle model attraction and repulsion force models are discussed.This solution has the major advantage over the existing scheduling solutions based on Ant or Particle Swarm optimization because it has a faster execution time, reduced complexity, it is distributed and highly resistant to disasters. Our solution tries to increase data locality, to reduce network traffic by data placement, tries to efficiently use the physical computing resources and offer good scheduling response time.