Third Generation Virtualization: Towards Applications’ Ubiquity

The cloud, which consists of data centers featuring a high concentration of CPUs and storage, is the natural habitat of the more popular Internet services, such as Facebook, Google Search, Google Drive, YouTube, etc. It is not limited to such large corporations: other companies can rent machines in the cloud to build innovative services. For example, Dropbox relies on Amazon Web Services to offer personal cloud storage services. SMEs and/or academic institutes can also rent clusters of machines to offload their work, e.g., big data analytics. The flexibility offered by virtualization is the key to the success of the cloud. Virtualization allows the share of computer hardware resources between multiple users or applications. The first-generation of virtualization, also known as heavy virtualization, consists in packaging one or several applications, alongside an entire operating systems (OS) such as Linux or Windows [1], into a so-called Virtual Machine (VM). An intermediate software layer, called the hypervisor, seats in between the VMs and the hardware and enables the correct operation of the VMs. The second-generation of virtualization, known as lightweight virtualization, is an emerging technology, led by the emblematic Docker platform1, where only the application is isolated within a so-called container. In this case, it is not necessary to embed the OS with the application [2], as all containers will share the OS of the physical machine. Heavy and light virtualization are more and more intertwined in public and private data centers (DCs). Indeed, heavy virtualization enables the decoupling from the hardware, while the light virtualization permits to pack ready-to-use applications.