Resource selection for fast large-scale Virtual Appliances Propagation

The increase of Dynamic Virtual Infrastructures usage brings up some problems. One of them is the efficient deployment, over large-scaled distributed systems, of the Virtual Appliances images. To address this problem two points needs to be faced, which nodes to select and how to transfer the VA images to those nodes. In this paper we propose a function for efficient node selection. This customizable function can be tailored to prioritize distribution time or node performance. We study how to tailor the function in order to balance both factors. We evaluate the performance of this function selection in conjunction with a deployment algorithm called Geometric Propagation obtaining exceptional results. We propose a mechanism that allows deploying a VA image over a large number of nodes within a reasonable period of time.