The 7 Things I Know About Cyber Security After 25 Years | April 2024
Application driven IT service management for energy efficiency
1. Application driven IT serviceApplication driven IT service
management for energymanagement for energy
efficiencyefficiency
Cinzia Cappiello, Alexandre Ferreira, Maria Grazia Fugini, Barbara
Pernici, Pierluigi Plebani
Dipartimento di Elettronica ed Informazione
Politecnico di Milano
[cappiello, ferreira, fugini, pernici, plebani]@elet.polimi.it
2. Green Information Systems
Energy-efficiency service centers
Energy-aware applications: annotations and Green Performance
Indicators
Conclusions and future work
OutlineOutline
2
3. ICT 4 Green:
◦ ICT used as a tool that enables energy savings techniques
Green IT
◦ ICT is considered as one of the field in which energy should be saved
◦ In the past, research focused on the energy efficiency at the
infrastructural level
Green ComputingGreen Computing
3
4. ICT and Energy ConsumptionICT and Energy Consumption
4
Data Centers
Home and Offices
Internet
6. Green IT drivers: Energy efficiencyGreen IT drivers: Energy efficiency
and energy leakage identificationand energy leakage identification
6
Energy leakage
highlight
the amount of
resources which are
not used towards the
achievement of the
business process goals
7. 7
Data Center energy efficiency:Data Center energy efficiency:
How is it possible to improve it?How is it possible to improve it?
Are the Servers always switched on?
Switch on/off servers when needed
Is the CPU (or other resources) scarcely used?
Virtualization and Consolidation
Voltage scaling
And for the storage?
Switch on/off disks
Slow down disks
8. Green Active Management of Energy in IT Service centres
◦ European Project fp7
The goal of GAMES is to develop methodologies, models, and tools
to reduce the environmental impact of information systems at all
levels, from application and services to physical machines and IT
plants.
Validation:
◦ Real data center
Expectation:
◦ Energy saving of 25%
Toward an adaptive data center:Toward an adaptive data center:
GAMES ProjectGAMES Project
8
9. From applications
To infrastructure
and back
GAMES focusGAMES focus
Energy-efficient service centersEnergy-efficient service centers
annotationsannotations
10. Monitoring Service center IS for assessment and
control
Applications
RQ 1: What is the optimum level
of information granularity of the
sensor network to optimize a
given flow network?
Sensors, energy consumption
models, aggregation of
monitored information
adaptivity actions in the service
center at different granularity
levels
Evaluation of Green
Performance Indicators (GPI) at
different levels of granularity
Adaptivity actions in the service
center at different granularity
levels
QoS requirements for
applications and their
components services, define
optional services
RQ 2: What information
granularity enables effective
enforcement of energy policy?
Design the minimal monitoring
infrastructure
Enable context-aware adaptivity
actions
dependencies among GPIs
strategical and tactical goals
Dependencies among GPIs Strategical and tactical goals
RQ 3: What information,
and at what level of
granularity, is required to
optimize a given type of flow
network?
Context-aware energy
management rules (at
application, platform,
infrastructure and facility levels)
RQ 4: What government policies
and regulations
will impel flow network managers
to make them
more energy efficient?
Eg apply EU code of conduct for
data centers
RQ 6: How can an
information system
integrate supply and
demand data to increase
energy efficiency?
Manage resources at a global
level, considering both
performance and energy
efficiency
Applications annotations to
characterize energy
consumption profile
RQ 9: What information do
consumers need
about the usage of the objects
they own or manage to increase
their
energy efficiency?
Needed resources for each
application
Requirements for an adaptiveRequirements for an adaptive
data centerdata center
10
13. As states before, in the past, research focused on the energy
efficiency at the infrastructural level
We focus also on the application level by addressing the following
issues:
◦ Estimation and assessment of the energy consumed by an application
◦ Identification of methods and actions for design applications that can be
made adaptive with respect to energy consumption and QoS
requirements
◦ Evaluation of the impact that the adaptive actions (and thus the saved
energy) have on the TCO of the IT Service Centre
Energy efficiency at the applicationEnergy efficiency at the application
levellevel
13
14. GAMES annotation for energy-GAMES annotation for energy-
aware applicationsaware applications
14
16. GPIs/KPIs Modeling (1/2)GPIs/KPIs Modeling (1/2)
16
CPU
Usage
CPU
Usage
CPU
Used
CPU
Used
CPU
Available
CPU
Available
DEFINITION OF OPERATIONAL DEPENDENCIES
GPI
Raw data
DC-EEPDC-EEP
IT-PEWIT-PEW SI-EERSI-EER
GPI
GPI
17. GPI1 is related to the GPI 2 by a
qualititative dependency if:
◦ Their trends are correlated
Qualitative dependencies can be
extracted by using data mining
tools
GPI/KPI modeling (2/2)GPI/KPI modeling (2/2)
17
DEFINITION OF QUALITATIVE DEPENDENCIES
GPI 1GPI 1
GP12GP12
18. The applications are enriched with specific
annotations at process level and activity level
GAMES-enabled applicationsGAMES-enabled applications
(1/2)(1/2)
18
O2O2
Attività 1Attività 1 Attività 2Attività 2
Attività 3Attività 3
Attività 4Attività 4
Application
Type:
business,
HPC, data
intensive
QoS
requirements:
e.g.,
Response time
<5s
QoS
requirements:
e.g.,
Response time
<1s
Control flow
property: e.g.,
mandatory vs
optional
O
1
,
s
e
q O2,
random
O2,
random
GPI: e.g. Transactions/KWh > 1000
20. In summary these annotations define the application in terms of:
◦ Type of the application (e.g., business, HPC, data intensive, ….)
◦ GPIs and KPIs for the application
◦ Data dependencies (e.g., data used by application, data dependencies
within activities in an application)
◦ Data characteristics
estimated I/O
access characteristics such as sequential or random
Volatility: permanent or transient information
variability in time
◦ Control flow properties (e.g., mandatory/optional activity, branch
execution probabilities)
◦ QoS properties at both activity and process level
◦ Energy consuming profile (from statistical and historical data)
◦ Data dependencies between processes
Process annotationsProcess annotations
20
22. Energy-aware adaptation(2/2)Energy-aware adaptation(2/2)
22
Energy- aware
Application
annotation
Energy- aware
Application
annotation
Additional
knowledge about
current and past
system
characteristics
Additional
knowledge about
current and past
system
characteristics
Suitable
adaptation
strategies
Suitable
adaptation
strategies
23. We have proposed an approach that enable evaluating energy
consumption in applications using annotations and GPI for energy-
awareness.
We are currently refining both the GPI by defining metrics for
resource (CPU, memory, disk, and so on) usage evaluation and
metrics regarding the application lifecycle, such as quality factors of
the employed development platform, data redundancies
We are developing a prototype that analyze data about application
executions to derive information about energy consumption and
identify resources responsible for energy leakages
Conclusions and future workConclusions and future work
23