Virtualizing Resources for the Cloud - Large Scale and Big Data: Processing and Management - page 498

Production virtual machines

A VM that can be used as a sandbox for:

Safely allowing attacks
that can be monitored so
as to develop defense against
them on the production VMs.
Cloning real VMs so as to
inspect incoming packets/input
before being forwarded to the
production VMs.

VM1 VM2 VM3 VM4

Virtualization intermediary (i.e., hypervisor or VMM)

Hardware

FIGURE 16.2

Using Virtual Sandboxes to develop defenses against attacks and to monitor

incoming data.

OSs, however, provides only process abstractions (not systems) that can access
resources in parallel. Virtualization takes this step forward by creating an illusion
of complete systems, whereby multiple VMs can be supported simultaneously, each
running its own system image (e.g., OS) and associated applications. For instance, in
virtualized datacenters, seven or more VMs can be provisioned on a single server,
providing potentially resource utilization rates of 60% to 80% [8]. In contrast, only
5% to 10% resource utilization rates are accomplished in nonvirtualized datacen-
ters [8]. By enabling multiple VMs on a single physical server, virtualization allows
consolidating physical servers into virtual servers that run on many fewer physical
servers (a concept referred to as server consolidation ). Clearly, this can lead not only
to improved system utilization but also to reduced costs.
Server consolidation as provided by virtualization leads not only to improved
system utilization and reduced costs but further to optimized energy consumption
in cloud datacenters. Datacenters hosting cloud applications consume tremendous
amounts of energy, resulting in high operational costs and carbon dioxide emissions
[11]. Server consolidation is perceived as an effective way to improve the energy
efficiency of datacenters via consolidating applications running on multiple physical
servers into fewer virtual servers. Idle physical servers can subsequently be switched
off so as to decrease energy consumption. Studies show that server consolidation can
save up to 20% of datacenter energy consumption [31,53]. A large body of research
work illustrates the promise of virtualization in reducing energy consumption in
cloud datacenters (e.g., [11-13,31,33,57]). Indeed, mitigating the explosive energy
consumption of cloud datacenters is currently deemed as one of the key challenges
in cloud computing.

16.1.5 F aCilitating b ig D ata a nalytiCs
The rapidly expanding Information and Communications Technology (ICT) that is
permeating all aspects of modern life has led to a massive explosion of data over the
last few decades. Major advances in connectivity and digitization of information