Proactive and reactive thermal optimization techniques to improve energy efficiency in data centers

Marina Zapater | Workshop PICATA | 14-‐02-‐2013 1
MONCLOA Campus of Interna4onal
Excellence
Proactive and Reactive Thermal
Optimization Techniques to Improve Energy
Efficiency in Data Centers
Workshop PICATA
Marina Zapater
José L. Ayala, José M. Moya

View Slide

Laboratorio de Sistemas Integrados (LSI)
Departamento de Ingeniería Electrónica
ETSI Telecomunicación
Universidad Politécnica de Madrid
ArTeCS Group
Group of Architecture and Technology of CompuQng Systems
Facultad de InformáQca
Universidad Complutense de Madrid
Presentation

View Slide

Data Centers

View Slide

Motivation
•  Power consumpQon in data centers
–  1.3% world energy producQon in 2010
–  USA: 80 billion KWh/year in 2011 = 1.5xNYC
–  250 billion KWh/year in 2010
•  More than 43 Million tons of CO2
/ year
•  More water than paper, automoQve, petrol, wood
or plasQc industry
Jonathan Koomey. 2011. Growth in Data center electricity use 2005 to 2010

View Slide

Motivation
•  It is expected for total data
center electricity use to
exceed 400 GWh/year by
2015.
•  The required energy for
cooling will conQnue to be at
least as important as the
energy required for the
computaQon.
•  Energy op4miza4on of
future data centers will
require a global and mulQ-‐
disciplinary approach.
0
5000
10000
15000
20000
25000
30000
35000
2000 2005 2010
High-‐end servers
Mid-‐range servers
Volume servers
0
50
100
150
200
250
300
2000 2005 2010
Infrastructure
CommunicaQons
Storage
High-‐end servers
Mid-‐range servers
Volume servers
5,75 Million new servers per year
10% unused servers (CO2
emissions
similar to 6,5 million cars)
World server installed base (thousands)
Electricity Use (billion KWh/year)

View Slide

State of the Art Energy Savings
for different abstraction levels
AbstracQon level
•  Higher levels of
abstracQon bring
more beneﬁts
•  ApplicaQon-‐level
sQll has to be
explored.
SoluQons proposed by the State of the Art
6
Marina Zapater | Workshop PICATA | 14-‐02-‐2013

View Slide

Our perspective

•  Using the knowledge about the energy demand of the
applica4ons, the features of the computa4on and
cooling resources to apply proacQve opQmizaQon
techniques
•  Global strategy to integrate mulQple informaQon
sources and coordinate decissions to reduce overall
power consumpQon.
Proactive and reactive
holistic approach

View Slide

Our perspective
•  State of the Art: PUE ≈ 1,2
– The important part is IT energy consumpQon
– Current work is focused on decreasing PUE
IT and Cooling power
cooling
IT
TOTAL
P
P
P +
=
IT
TOTAL
P
P
PUE =

View Slide

Our perspective
IT and Cooling power
cooling
IT
TOTAL
P
P
P +
=
IT
TOTAL
P
P
PUE =
•  Minimize IT power
•  Jointly minimize IT and cooling

View Slide

Minimizing IT Power
Leveraging heterogeneity
•  Usage heterogeneity (existance of diﬀerent servers) to
minimize energy consumpQon:
–  StaQc: Finding the best data center set-‐up, given a number of
heterogeneous machines
–  Dynamic: opQmizaQon of task allocaQon
M. Zapater, J.M. Moya,
J.L. Ayala. Leveraging
Heterogeneity for
Energy MinimizaQon in
Data Centers, CCGrid
2012
10

View Slide

Minimizing IT power
Application Awareness
WORKLOAD
Scheduler Resource
Manager
Execu4on
11

View Slide

Heterogeneity
Application Awareness
WORKLOAD
Scheduler Resource
Manager
Execu4on
Proﬁling and
Classiﬁca4on
Energy
Op4miza4on
12

View Slide

Cooling management
•  Control of the fan speed of a
server
–  Enterprise server: Sparc T3 (256
threads)
–  Real measures with server
internal sensors
•  We can ﬁnd an opQmum
pointbetween leakage and
cooling to minimize power
Work in collaboraQon with:

Leakage-cooling tradeoffs at the
server level

View Slide

Cooling management
Fig. 4. Test 3 temperature sensor readings for the three different controlle
0.1
0.2
)
Energy difference between 1800RPM and 2400RPM for clustered allocation
analytical model for leakag
fan speeds for varying utiliza
4. Test 3 temperature sensor readings for the three different controllers
0RPM for clustered allocation
analytical model for leakage power and ﬁnd the optimum
fan speeds for varying utilization values. Based our analytical
Work in collaboraQon with:

Leakage-cooling tradeoffs at the
server level
14

View Slide

Cooling & IT Joint Opt.
•  Deriving a data room thermal
model to jointly allocate
computaQonal and cooling
resources
–  Gathering environmental data
through sensors (WSN)
–  Server sensors
–  Workload informaQon
•  Usage of geneQc programming
and geneQc algoQthms
Work in Progress

View Slide

Holistic aproach
•  System that increases the knowledge of the data center
•  Real implementaQon scenario at CeSViMa
•  Power opQmizaQon in globally distributed systems

Proactive and reactive techniques
Data Center
state Op4miza4on
Datacenter
Sensing
Decission
proposal
GreenDISC Project: HW/SW
Technologies for Energy
Eﬃciency in Distributed
Compu4ng Systems.
UCM-‐UPM

TEC2012-‐33892.
Spanish Ministry of Economy and
CompeQQveness

View Slide

Research goals
•  Energy and CO2 carbon footprint reducQon in
data centers
– So far, 25% reducQon in IT resource management
opQmizaQons
– 10% energy reducQon in fan control policies
•  Joint IT/cooling techniques are expected to
bring much more beneﬁts.
•  SoluQons in a real environment: CeSViMa

Expected impact

View Slide

Questions?
Thank you for
your a^en4on
Marina Zapater
[email protected]
B105. ETSI Telecomunicación
Avda Complutense, 30
91 549 57 00 (+ 4227)
hsp://greenlsi.die.upm.es
hsp://artecs.dacya.ucm.es/

View Slide

Research results
•  M. Zapater, J.L. Ayala, J.M. Moya, K. Vaidyanathan, K. Gross, A.K. Coskun, Leakage and Temperature
Aware Server Control for Improving Energy Eﬃciency in Data Centers. To apper in: DATE’13, 2013.
•  M. Zapater, J. L. Ayala, and J. M. Moya. “GreenDisc: a HW/SW energy opQmizaQon framework in globally
distributed computaQon”, J. Bravo, D. López-‐de Ipiña, and F. Moya, Ed., Springer Berlin Heidelberg, 2012,
pp. 1-‐8.
•  M. Zapater, J. L. Ayala, and J. M. Moya, “Leveraging heterogeneity for energy minimizaQon in data
centers”, in Proceedings of the 2012 12th IEEE InternaQonal Symposium on Cluster, Cloud and Grid
CompuQng (CCGRID 2012), Washington, DC, USA, 2012.
•  M. Zapater, C. Sanchez, J. L. Ayala, J. M. Moya, and J. L. Risco-‐Maryn, “Ubiquitous green compuQng
techniques for high demand applicaQons in smart environments” Sensors, vol. 12, iss. 8, pp. 10659-‐10677,
2012
•  M. Zapater, P. Arroba, J. M. Moya, and Z. Bankovic, “A State-‐of-‐the-‐Art on energy eﬃciency in today’s
datacentres: researcher’s contribuQons and pracQcal approaches”, UPGRADE, vol. 12, iss. 4, pp. 67-‐74,
2011.
•  M. Zapater, J. L. Risco, J. L. Ayala, and J. M. Moya, “Combined Dynamic-‐StaQc approach for Thermal-‐
Awareness in heterogeneous data centers” IWIA 2010.

View Slide

Proactive and reactive thermal optimization techniques to improve energy efficiency in data centers

Proactive and reactive thermal optimization techniques to improve energy efficiency in data centers

GreenLSI

More Decks by GreenLSI

Other Decks in Research

Featured

Transcript