Information Technology Reference
In-Depth Information
Cloud Computing can be viewed as a combination of many preexisting tech-
nologies such as virtualization and server consolidation, among others.
The term virtualization refers to the abstraction of compute resources (CPU,
memory, I/O) from the applications, aiming to improve sharing and utilization
of computer systems. One immediate benefit of virtualization is the option to
run multiple operating systems and software stacks on a single physical platform.
Server consolidation is the process of gathering several Virtual Machines
(VMs) into a single physical server. It is often used by Data Centers to increase
resource utilization and reduce electric power consumption costs [11]. For exam-
ple, consider a set of VMs
,
each of them with a quad-core processor, where each processor is capable of ex-
ecuting one VM. A power ecient allocation schedule could initially assign all
the VMs to the same host in such a way that the other hosts could be put in
the power-saving state or turned off. A possible solution to this problem would
be, therefore, to pack the maximum workload in the smallest number of servers,
keeping each resource (CPU, disk, network, among others) on every server at
100% utilization and put the idle servers in power-saving state.
The consolidation process can be performed in a single step using the peak
load demands, known as static consolidation, or in a dynamic manner, by reeval-
uating periodically the workload demand in each VM. In static consolidation,
VMs stay in the same physical server during their whole lifetime. The utilization
of the peak load demand should ensure that the VM does not overload. How-
ever, in a dynamic environment with different access patterns, one or more VMes
can become idle, resulting in an inecient power allocation.Dynamic consolida-
tion aims to tackle this problem by taking into account the current workload
demands. Dynamic consolidation may require migrating VMs between physical
servers in order to [7]: (i) pull out physical servers from an overload state or (ii)
turn off a physical server when it is idle or when the VMs mapped to it can be
moved to another physical server.
Server consolidation in a Cloud Computing environment presents some ad-
ditional diculties since the Cloud must also provide reliable QoS, normally
defined in terms of Service Level Agreements (SLA). An SLA describes charac-
teristics such as maximum throughput and minimum response time, that must
be delivered by the deployed systems. If an SLA is violated, economical penalties
usually apply.
{vm 1 ,vm 2 ,vm 3 ,vm 4 }
and a set of hosts
{h 1 ,h 2 ,h 3 }
3 Energy-Aware Computing
Cloud Computing solutions may have a potential impact on green house gas
(GHC), which include CO 2 emissions. Saving energy of a Data Center with ac-
ceptable QoS requirements is an economical incentive for data center operators,
as well as a significant contribution to the environment. This requires the de-
sign of energy-aware solutions. Energy-awareness can be characterized by taking
into account the amount of resources and QoS requirements required by the
applications and also the energy requirements along their life cycle [9].
 
Search WWH ::




Custom Search