Information Technology Reference
In-Depth Information
mainframes. But due to low speed of mainframes at that time (by slow, we
are talking about IBM System 370) and the later introduction of powerful
operating systems like UNIX, which had a strong capability of resource
isolation between users, the research and development of virtual machines
were slowed down. Virtualization was effectively abandoned during the
1980s and 1990s when client-server applications and inexpensive x86 serv-
ers and desktops established the model of distributed computing [2]. But
the development of grid computing, in addition to other forms of Internet-
based computing, raises some important operational challenges like low
infrastructure utilization, increasing physical infrastructure cost, increas-
ing IT management cost, security, and insufi cient failover and disaster
protection [2]. VM technology [4] has recently been studied extensively to
show its ability in supporting system-level virtualization of computing
environments to solve the above-mentioned problems of resource abstrac-
tion and better management. VM and virtual network (VN), generates a
virtual computing environment in which virtual resources are mapped to
physical resources. A virtual environment (VE), also known as a private
virtual grid or virtual private working space [5], has many advantages. It
allows users to work in a virtualized but familiar environment, provides
an extra layer of hardware isolation and security, defends computing
environments against possible attacks from malicious applications, and
protects the privacy of virtual environments. Owing to these advantages,
virtualization has regained its popularity in recent years. Virtualization
enables better resource management by marshalling users' coni guration
requests for physical resources. Although these works have shown the
potential in supporting system-level virtual environments, they have not
adequately addressed how to enhance the current VM technology so that
it can adapt to a grid or a general shared cyberspace. The major problem
is that in a general shared and distributed environment, resources are
heterogeneous and their availability varies with time. In addition, users'
requirements also change with time. These heterogeneity and dynamics
demand virtual environments supporting not only swift instantiation
but also the ability of reconi guration and migration from time to time to
adapt to the resource availability making conventional virtualization
technologies inadequate to provide an efi cient virtual environment (VE)
on grid systems. Coni guring an appropriate VE in a general parallel and
distributed environment is at least as complex as that of a physical system.
Furthermore, current virtualization platforms have mainly focused on
functionality and have limited support for coni guration management of
virtual environments.
System-level virtualization is not only critical for grid computing, but
also vital for the next generation network computing in general. Let us
consider today's distributed environment and the IT industry in general,
where computing paradigms such as service-oriented architecture, utility
computing, business-on-demand, adaptive enterprise, and data center
Search WWH ::
Custom Search