Scheduling of Large-scale Virtualized Infrastructures: Toward Cooperative Management

This book addresses software managers dedicated to distributed virtual infrastructures. It focuses especially on how to leverage system virtualization to schedule dynamically and reactively a huge number of jobs.

Over the past few years, system virtualization has been increasingly used in distributed systems because of the advantages it offers to the owners and users of these infrastructures. This has led to the development of software managers in charge of operating these infrastructures. Currently, most of these managers are highly centralized, and this restricts their scalability. In this book, we study the management of large-scale virtual infrastructures. We focus in particular on how to schedule dynamically and reactively up to 80,000 virtual machines on 8,000 nodes.

Nowadays, virtual infrastructures tend to be bigger and bigger, including up to 80,000 nodes. However, most of the existing software managers are not designed to handle the problems appearing at this scale. Therefore, new approaches need to be investigated to meet the evolutions of virtual infrastructures. This is exactly the goal of this book (which focuses on the scheduling of virtual machines) that is the first one to study this topic, as far as we know.