IEEE Access,
Journal Year:
2024,
Volume and Issue:
12, P. 105234 - 105245
Published: Jan. 1, 2024
The
dynamic
landscape
of
cloud
computing
design
presents
significant
challenges
regarding
power
consumption
and
quality
service
(QoS).
Virtual
machine
(VM)
consolidation
is
essential
for
reducing
usage
enhancing
QoS
by
relocating
VMs
between
hosts.
OpenStack
Neat,
a
leading
framework
VM
consolidation,
employs
the
Modified
Best-Fit
Decreasing
(MBFD)
placement
technique,
which
faces
issues
related
to
energy
QoS.
To
address
these
issues,
we
propose
an
Energy
Efficient
Consolidation
(EEVMC)
approach.
Our
method
introduces
novel
host
selection
criterion
based
on
incurred
loss
during
identify
most
efficient
host.
For
validation,
conducted
simulations
using
real-time
workload
traces
from
Planet-Lab
Materna
over
ten
days,
leveraging
latest
CloudSim
toolkit
compare
our
approach
with
state-of-the-art
techniques.
Planet-Lab's
workload,
EEVMC
shows
reduction
in
80.35%,
59.76%,
21.59%,
7.40%,
fewer
system-level
agreement
(SLA)
violations
94.51%,
94.85%,
47.17%,
17.78%
when
compared
(MBFD),
Power-Aware
Best
Fit
(PABFD),
Medium
Power
(MFPED),
Power-Efficient
(PEBFD),
respectively.
Similarly,
Materna,
achieves
16.10%,
61.0%,
4.94%,
4.82%,
SLA
76.99%,
88.88%,
12.50%,
48.65%
against
same
benchmarks.
Additionally,
Loss-Aware
Performance
(LAPED)
significantly
reduces
total
number
migrations
time
per
active
host,
indicating
substantial
improvement
efficiency.
Concurrency and Computation Practice and Experience,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Sept. 30, 2024
Summary
Cloud
data
centers
(CDCs)
have
revolutionized
global
computing
by
offering
extensive
storage
and
processing
capabilities.
Nevertheless,
the
environmental
impact
of
these
processes,
including
their
substantial
energy
consumption
carbon
emissions,
calls
for
implementing
more
efficient
techniques.
Efficient
virtual
machine
(VM)
consolidation
is
crucial
in
optimizing
resource
utilization
reducing
consumption.
Current
methods
enhancing
efficiency
often
lead
to
issues
such
as
service
level
agreements
(SLAs)
violations
quality
services
(QoS)
degradation.
This
study
presents
a
novel
approach
host
selection
using
grey‐extreme
(GE)
learning
model,
which
accurately
predicts
over
underutilized
hosts.
In
addition,
VM
placement
technique
called
enhanced
black
widow
optimization
(EBWO)
utilizes
heuristic
techniques
differential
evolutionary
optimize
placement.
The
proposed
dynamic
optimizes
while
meeting
strict
SLA
requirements
QoS
metrics
CDCs.
Extensive
analyses
were
conducted
Cloudsim
toolkit
validate
approach's
effectiveness.
These
encompassed
conditions
random
workloads
heterogeneous
environments.
simulation
results
showed
that
GE‐EBWO
outperforms
other
improves
12%–15%.
it
significantly
decreases
migrations
11%–14%
compared
advanced
methods.
validates
practicality
moving
towards
environmentally
friendly
Practice, progress, and proficiency in sustainability,
Journal Year:
2024,
Volume and Issue:
unknown, P. 56 - 73
Published: Feb. 7, 2024
Green
computing
can
be
described
as
efficient
resource
management
in
distributed
environments
such
cloud,
fog,
and
edge
environments.
In
green
computing,
refers
to
the
eco-friendly
environment
with
environmental
responsibility
efficiently
manage
resources.
Mainly,
two
significant
reasons
are
associated
global
warming
issues
from
perspective.
They
high
power
consumption
of
cloud
datacenters
CO2
emission
rate.
According
many
survey
reports,
every
year,
alone
produce
nearly
90
million
metric
tons
into
environment,
now,
this
has
become
one
primary
for
issues.
Also,
it
stated
that
rate
would
increase
by
8%
year
if
did
not
identify
proper
control
measurements.
Therefore,
is
crucial
enhance
e-waste
increasing
efficiency
usage
minimizing
consumption,
rate,
inefficient
recycling
policies,
etc.
Businesses
must
outbid
suppliers
to
win
over
new
clients
in
a
very
competitive
industry.
Since
client
retention
directly
affects
company's
revenue,
it
is
prominent
topic
for
research.
can
prevent
attrition
by
proactively
addressing
when
occurs.
A
two
stage
predictive
ensemble
method
proposed
with
Gaussian
Naive
Bayes,
Perceptron,
Gradient
Boosting
Classifier,
Decision
Tree,
K-Nearest
Neighbors,
Logistic
Regression,
and
XGBoost
used
the
first
model
classification,
then
models
like
Adaboost,
Bagging
Stacking,
Voting
are
second
increase
forecast
accuracy
top
four
chosen
stage.
The
LIME
Explainer
algorithm
also
analyzes
data
instance
adjustments
their
effect
on
predictions,
improving
interpretability
"black
box"
models.
This
two-step
approach
addresses
issues
crucial
decision-making,
optimizing
final
model's
power
interpretability.
This
paper
proposes
an
Enhanced
Virtualization
of
Resources
(EVR)
system
for
high
performance
applications
in
Cloud
Computing.
It
uses
a
Deep
Regression
Model
(DRM)
to
predict
the
resource
requirements
application
be
deployed
on
Cloud.
The
model
takes
into
account
various
parameters
like
number
users,
bandwidth
requirements,
processing
time,
I/O
requests
and
server
capability
make
accurate
predictions.
is
further
optimized
with
Genetic
Algorithm,
which
mutation,
crossover
selection
operations
ensure
produces
high-accuracy
output.
resulting
then
used
by
EVR
decide
nodes
should
allocated
best
performance.
evaluated
using
metrics
such
as
query
response
allocation
accuracy.
Results
demonstrate
that
proposed
can
provide
up
73.3%
more
efficiency
than
existing
approaches
virtualization.
Управление большими системами.,
Journal Year:
2024,
Volume and Issue:
109, P. 268 - 292
Published: May 31, 2024
В
связи
с
возрастающим
спросом
на
облачные
сервисы
разработка
новых
эффективных
методов
и
алгоритмов
планирования
ресурсов
облачных
центров
обработки
данных
является
актуальной
задачей,
о
чем
свидетельствует
непрекращающийся
поток
работ,
посвященных
данной
теме.
Критерии
лучшего
распределения
могу
быть
различными,
такие
как
энергоэффективность,
выполнение
соглашений
об
уровне
сервиса,
надёжность
другие.
На
основе
проанализированных
работ
были
выбраны
разработаны
модели,
методы
алгоритмы
ресурсов,
комплекс
которых
положен
в
основу
предложенной
статье
архитектуры
распределенного
планировщика
инфокоммуникационной
системы
облачного
ЦОД
многокритериальной
оптимизации
ее
характеристик
особенностей
живой
миграции
виртуальных
машин.
Эффективность
использованных
моделей
подтверждена
имитационным
моделированием.
Показано,
что
предложенный
позволяет
сократить
энергопотребление
при
выполнении
показателей
качества
обслуживания.
With
the
development
of
cloud
technologies,
methods
and
algorithms
for
a
resource
scheduler
data
centers
is
an
urgent
task,
as
evidenced
by
continuous
flow
works
devoted
to
this
topic.
The
criteria
best
allocation
resources
can
be
different,
such
energy
efficiency,
fulfillment
service
level
agreements,
reliability
others.
Based
on
analyzed
works,
models,
distribution
were
selected
developed,
complex
which
forms
basis
distributed
architecture
proposed
in
article
infocommunication
system
center
based
multi-criteria
optimization
its
characteristics
features
live
migration
virtual
machines.
effectiveness
models
used
has
been
confirmed
simulation
modeling
makes
it
possible
reduce
consumption
when
meeting
quality
indicators.
