AERTEC - 2009

IBM Research
Innovative Data Center Energy
Efficiency Solutions
Dr. Hendrik F. Hamann
IBM T.J. Watson Research Center
2/8/2009
© 2007 IBM Corporation
IBM Research
A holistic Challenge: Energy & Thermal Management
• Energy / thermal management is relevant on all levels
• Various length and times scale and interdependencies are involved
but also many analogies/similarities exist
• Truly holistic understanding is required to conquer the challenge
2
AERTC Conference
© 2008 IBM Corporation
IBM Research
Glo
ba
Tem l
per
at
000
km
yea
rs
Con
ti
Pow nenta
l
erN
ets
100
40’
5’ 0
00
km
Yea
r
100
k
Day m
Pow
e
Sta r
tio
n
ure
A broader Perspective
• The challenge is even bigger: Energy/thermal issues propagate all the way to the
world climate
• Earth has an energy and thermal problem as well
3
AERTC Conference
© 2008 IBM Corporation
IBM Research
Thermal Management and Hotspots
• Hotspots exist on all levels
• Cooling hotspots cost (a lot of) energy and determine cooling energy efficiencies
• …but opportunities for mitigation exist
(i.e., static, dynamic, spatial, temporal, spatial-temporal)
4
AERTC Conference
© 2008 IBM Corporation
IBM Research
Thermal Management and Hotspots
Microprocessor
~ 300 M transistors
Data Center
~ 1000 of servers
US Power Grid
~ 300 M customers
Superstore / Airports
~ 1000 of customers
CRAC units
server
racks
perforated
tiles
under floor
plenum
5
AERTC Conference
© 2008 IBM Corporation
IBM Research
Data Center Facts
• DCs consume ~ 2 % of all US electricity
• annual growth (15 %) is non-sustainable
• DC power projected to be > 8 % of US
power by 2020
• governments consider regulatory actions
• every DC is different, DCs are
heterogeneous and change over time
• DCs are not as efficient as they should
• inefficiencies are caused by lack of
best practices
• best practices are hard to manage
because they are hard to measure
6
AERTC Conference
© 2008 IBM Corporation
IBM Research
Content
How to measure, model and manage data center energy
efficiency ?
 DC energy efficiency: PUE and beyond
 from a Mobile Measurement Technology (MMT 1.0)….
 need for spatially dense data
 a first solution
 case study and results
 to a Measurement Management Technology (MMT 1.5)…
from a static to a dynamic solution
 energy and thermal models
 case study and results

7
AERTC Conference
© 2008 IBM Corporation
IBM Research
Data Center Energy Efficiency – PUE Metric
 PUE is widely used today: PUE = Total DC Power / IT Power
 many PUE “claims” – but PUE metric can be problematic
 weather-dependent, location dependent, application/tier dependent
 non-linear, awards UPS consumption, power density dependent
 PUE does not include IT performance
 PUE metering is often not in place
 PUE is often insufficient for “proving” and managing energy efficiency
8
AERTC Conference
© 2008 IBM Corporation
IBM Research
A more detailed Look – DC Energy Efficiency
Transport COP
COPtrans  PRF /
Thermodynamic COP
*  PRF /PChiller
COPthermo
• Average Chiller COP
• (throughout the year)
AERTC Conference

i 1
i
PACU
Data Center COP
1/ COP  1/ COPthermo  1/ COPtrans
THERMODYNAMIC PART OF COOLING:
HOTSPOTS / HIGH INLET TEMPERATURES
IMPACT CHILLER EFFICIENCY (~ 1.7 % per F)
9
# of active ACUs
TRANSPORT PART OF COOLING:
LOW ACU UTILIZATION IMPACTS ACU
BLOWER CONSUMPTION (~ 5-8 kW/ACU)
© 2008 IBM Corporation
IBM Research
Visualizing, Measuring and Managing Data Center Best Practices
Mobile Measurement Technology
Temperature
 designed to optimize DC resources to reduce up to 15% of DC energy consumption
scans, digitize rapidly physical environment (temperature, flow, pressure etc..) of DC
 cart tool comprises sensor network, where each sensor defines a virtual unit cell
 technology is based on interworking between measurements, models and DC management
10
Hot spot
@
AERTC Conference
5.5
t
f ee
Hot air mixing
with cold air
© 2008 IBM Corporation
IBM Research
IBM Mobile Measurement Technology (MMT 1.0)
Solution Approach – Three Steps
1
2
Measure
Model
 Capture high
resolution
temperature
data, air flow
data and
infrastructure &
layout data
3
 To identify
improvement
opportunities
model the data
center and use
optimization
algorithms (“best
practices rules”)
Manage “Best Practices”
 Realize air transport energy savings
 Realize thermodynamic energy savings
 Achieve reduced energy consumption
 Potential for deferring new investments
11
AERTC Conference
© 2008 IBM Corporation
IBM Research
MMT 1.0 @ Work – 3D Heat Maps
MMT – Scans:
Thermal measurements
at different heights
(1 ft increments in z)
max
 detailed 3D heat maps
(<40 mins scan time)
 30000 thermal readings
 3000 humidity readings
 200 air flow sensor
min
12
AERTC Conference
© 2008 IBM Corporation
IBM Research
MMT 1.0 @ Work – Energy Savings
Case Study: DC Area = 20k sqf; Temp. Meas. = 200,000; Airflow Meas. = 1,200; Power density ~ 75 W / sqf
Thermo Savings
BEFORE
34 % =120 kW
= 20 kW
BEFORE
AFTER
Transport
Savings
AFTER
Thermo
Savings
7 F=37 kW
Increase Chiller Set-Point
13
AERTC Conference
Increase ACU Utilization
© 2008 IBM Corporation
IBM Research
Typical Energy Savings
Saving = 177 kW
Cool
power consumption
 saved 177 kW with measurement / metrics driven
best practices implementation
 developed 6 tier metric to drive best practices
implementation with minimal investments
 typical 1-2 Month turnaround to realize savings
 Improved DC COP 2.39 to 3.44
 COPthermo from 4.5 to 5.1
 COPtrans from 5.3 to 9.8
IT
before
Finding / Metrics
Key Action / Solution
Horizontal hotspots (HH)
change tile layout & deploy high throughput tiles
Vertical hotspots (VH)
snorkels / fillers
Non-targeted air flow
close leaks / cable cutouts
Plenum temperatures
service ACUs supply side / increase ACU utilization
ACU utilization
turn under-utilized ACUs off
ACU flow
remove blockage
Cool
IT
after
thermo
transport
 Case Study: DC Area = 20k sqf; Temp. Meas. = 200,000; Airflow Meas. = 1,200; Power density ~ 75 W / sqf
14
AERTC Conference
© 2008 IBM Corporation
IBM Research
MMT 1.0 - Status
 MMT service provided to more than 30 DCs
(different sizes, power densities, locations etc.)
 repeatedly identified energy savings of > 10 % of IT power
(to date more than 35 M kW hours)
 MMT has delayed major DC upgrades / capital investments
 MMT is being deployed in all IBM’s strategic DCs in NA
(saving target of more than 17 M kW hours)
 MMT 1.0 is a service offering in 3 GEOs (NA, EMEA, AP,…)
15
AERTC Conference
© 2008 IBM Corporation
IBM Research
MMT 1.5 -From a static to a dynamic Solution
max
 DC can change over time
 IT power levels can change (e.g., 10-15 % during a day)
 cooling conditions change etc..
 new racks / new servers / re-arrangement of tiles etc..
min
 MMT 1.0 is “sparse” in time but “dense” in space
 Real-time sensor are “sparse” in space but dense in time
 MMT 1.5 provides high time & spatial resolution combining
 MMT 1.0 for base model generation, sensor placement etc..
 real-time sensors for creating dynamic models
Animation of 3D heat map
over 24 hours
Real time Sensors:
+
16
AERTC Conference
+
© 2008 IBM Corporation
IBM Research
MMT 1.5 – Measurement & Management Technology
Evolution from MMT 1.0 to MMT 1.5
MMT 1.0: Dense in Space
MMT 1.0: Detailed Report
MMT 1.5: Dense in Time
MMT 1.5: DC Management Solution
MMT 1.0: Detailed Base DC Model
MMT 1.5: Dynamic DC Model
17
AERTC Conference
© 2008 IBM Corporation
IBM Research
Summary
• MMT 1.0 has repeatedly shown energy efficiency
improvements by more than 10 %
http://www.youtube.com/watch?v=feF7vFj4Deo
• MMT is being extended to an active energy management
energy solution by combining MMT models with real-time
sensor data (MMT 1.5)
• MMT leverages different models based on data
availability, and application
18
AERTC Conference
© 2008 IBM Corporation
Similar pages