Noise within a Data Center

Abstract

Data centers with its numerous servers, network switches, routers and air conditioning equipment produce significant noise that influences work and communication of maintenance staff. Noise levels and spectrum within a data center are measured and analyzed. Daily and weekly variations of noise levels are also captured. From measured noise levels and spectrum speech intelligibility measures are calculated and communication distances within a data center are determined. The need for noise protection and methods for noise reduction are considered.

Full text

1352 MIPRO 2016/CTS
Noise within a Data Center
Dubravko Miljković
Hrvatska elektroprivreda, Zagreb, Croatia
dubravko.miljkovic@hep.hr
Abstract - Data centers with its numerous servers, network
switches, routers and air conditioning equipment produce
significant noise that influences work and communication of
maintenance staff. Noise levels and spectrum within a data
center are measured and analyzed. Daily and weekly
variations of noise levels are also captured. From measured
noise levels and spectrum speech intelligibility measures are
calculated and communication distances within a data
center are determined. The need for noise protection and
methods for noise reduction are considered.
I. INTRODUCTION
Work in a data center exposes staff to some
suboptimal working conditions. Noise, temperature and
air circulation are three of the biggest environmental
issues that data center workers face. Air-condition and
equipment cooling fans necessary for the proper operation
of IT equipment run continuously and create excessive
noise that influence comfort, poses risk to hearing and
impairs communication and concentration. With the
introduction of smaller hardware, data center is even more
densely populated and noise inside is becoming louder.
Constant humming of air-conditioners and servers within
a data center may create noise sometimes in excess of 80
dBA. Although maximum acceptable 8 hour exposure
level in most countries is set at 85 dBA, as 10% of the
level that does not show any effect (NOEL - No Observed
Effect Level), this level is far above levels prescribed for
mentally demanding work and office environment.
II. SOURCES OF DATA CENTER NOISE
Data centers consist of servers, routers, switches,
storages, tape systems, UPS and cooling equipment
(HVAC – Heat Ventilation Air Condition). Main sources
of data center noise are shown in Fig. 1. HVAC with its
supply ventilation ducts is a single main source of a data
center noise, with noise level typically about 70 dBA.
Within IT equipment main contributors to data center
noise are cooling fans installed in numerous servers,
routers, switches and storages, although some minor noise
comes from disks, tape drives, transformers and human
Figure 1. Main sources of data center noise
intervention. Individual servers produce noise levels in a
40-70 dBA range, depending on the brand and size, [1].
The sound power (not SPL) of centrifugal and axial
fans (such are used in HVAC ventilation ducts) is given
by (1), [2]:
NWW CBFIPQKL 



1010 log20log10 (1)
where
LW sound power level
KW specific sound power level depending on a type
of fan (obtained from manufacturer data)
Q volume flow rate
P total pressure (inches of H2O)
BFI blade frequency increment, correction for pure
tone (obtained from the graphs)
CN efficiency correction (out of optimum flow
conditions, obtained in tables)
When there are hundreds of servers, with hundreds of
cooling fans, noise levels are considerably higher than in
the case of one server. Considering them as noise sources
that are not coherent, level difference is given by (2):
ΔL = 10 × log n (2)
where
ΔL is the level difference
n is the number of equally loud noise sources
Permissible Exposure Time for various noise levels
(according to NIOSH) may be calculated using (3), [3].
t = 480 / 2(L - 85)/3 (3)
where
t is maximum exposure duration (seconds)
L is noise exposure level (dBA)
3 is exchange rate (dBA), i.e. for noise level increase
of 3 dBA, permissible exposure time is cut in half
Permissible exposure time for noise - guidelines for
level and duration (time) are shown in Table I, [4]. Noise
exposure level/times exceeding those shown in table
require use of hearing protection. When considering
working conditions for technicians and visitors in a data
center, high noise levels influence not only hearing and
communication but negatively influence concentration
needed for performing complex cognitive tasks. Some
kind of protection may be needed, especially for workers
working long shifts to increase concentration, comfort and
TABLE I. PERMISSIBLE TIME EXPOSURE
Level, dBA 85 88 90 92 94 95 100 105 110 115
OSHA PEL 16 8 4 2 1 0,5 0,25
NIOSH REL 8 4 1 0,25
Duration (in hours) of allowable exposure based on OSHA and NIOSH
criteria. PEL = Permissible Exposure Limit; REL = Recommended
Exposure Limit

MIPRO 2016/CTS 1353
prevent undesirable extra-aural effects (influence on the
autonomous system), [5]. When noise levels exceed 85
dBA, hearing conservation programs are required. This
includes baseline audiometric testing, noise level
monitoring or dosimetry, noise hazard signage, education
and training, [6]. Once levels exceed 87 dBA (in Europe)
or 90 dBA (in the US), further measures are required like
use of hearing protection, rotation of employees, or
engineering controls must be taken, [6].
III. SPEECH COMMUNICATION UNDER NOISY
CONDITIONS
During various installation and maintenance works
communication between staff is of utter importance.
Normal conversation tends to occupy 60-70 dBA range.
Speech communication in data center is degraded by the
masking effect of the background noise and changes in
vocal effort are necessary for various background noise
levels, [7], as shown in Fig. 2, [8], and Table II.
Figure 2. Long-term spectrum of speech under various vocal efforts
(adopted from [8])
TABLE II. SPEECH LEVELS AT VARIOUS VOCAL EFFORTS
Voice Average level dB/dBA
Casual 52.0/42.0
Normal 57.0/47.0 (private speech)
Raised 64.0/57.0
Loud 73.0/62.0
Shout 85.0/72.0
Auditory masking is intrusion of unwanted sounds
that interfere with the speech signal. Masking effect of
data center noise in spectral domain is illustrated in Fig.
3-5. When the low-frequency noise is louder than the
Figure 3. Spectrogram of speech (‘Noise within a data center’)
Figure 4. Spectrogram of data center noise
Figure 5. Spectrogram speech plus data center noise combined
speech signal it effectively masks speech. At high sound
pressure levels such noise effectively masks both vowels
and consonants. High-frequency noise masks only the
consonants, and its masking effectiveness decreases as
the noise gets louder, [9]. Noise exposure levels in a data
center afford less than desired intelligibility (< 95%).
IV. DATA CENTER NOISE MEASUREMENTS
Noise measurements were performed in mid-size
corporate data center, of surface area about 170 m2. IT
equipment installed within a data center is listed in Table
III. Temperature inside a data center is kept in 21-22 °C
range with separate cabinet cooling systems blowing out
air at 18-19 °C.
TABLE III. IT EQUIPMENT WITHIN A DATA CENTER
Type Number of devices
Server (physical) 120
Storage 5
Tape subsystem 3
Routers/ Switches 15
Cabinet Cooling System 10
A. Measuring Equipment
Noise levels were measured using the CEM DR-805
Sound Meter. A-level weighting was used due to high
correlation with people’s subjective judgment of the
loudness. Noise signals were recorded at 44.1 kHz
sampling frequency and 16-bit resolution, using simple
Ednet Desktop Microphone with 30 Hz-16 kHz range, HP
ProBook 6570b laptop computer with internal audio card
and YMEC sound measurement and analyzing software.

1354 MIPRO 2016/CTS
Figure 6. Data center schematic
B. Measured Noise Levels
Noise was measured at 19 different places, 15 in a
data center and 4 in accompanying spaces (console room,
archive, operators room and at the entrance hall) shown
in data center schematic, Fig. 6. All noise measurements
were performed at the head level while in standing
position. Measurement results are shown in Table IV.
Noise level depends if it is measured in a cold or hot
aisle. The lowest level in a data center was around 70
dBA and the highest slightly above 80 dBA. Similar
values are common in data centers, [10]. Higher levels
were measured in hot aisles (where cooling air gets out of
servers). The highest level was measured near routers
with particularly noisy cooling fans. These levels are
bellow maximum acceptable levels. However repeated
exposure to noise levels between 75 and 85 dBA may
pose a small risk to some people, [11]. Being so close to
the limit and considering excess risk of long term
exposure, [4], it may be wise to take no chances and wear
some hearing protection. Noise in accompanying spaces
was much lower (under 60 dBA). Although this is much
less than in data center and staff got used to it, prolonged
stay is still quite unpleasant for unaccustomed visitors.
C. Noise Spectrum
Spectrum of data center noise is shown in Fig. 7.
Please note harmonic content at frequency range from
150 Hz up to 1.5 kHz that originates from air condition
TABLE IV. NOISE LEVELS AT VARIOUS POSITIONS
Position Description, contributing sources in aisle dBA
1 Servers, storage, cabinet cooling system 76,2
2 Servers, storage, cabinet cooling system 76,1
3 Servers, storage, cabinet cooling system 79,2
4 Servers, storage, cabinet cooling system 76,9
5 Servers, storage, cabinet cooling system 78,1
6 Servers, cabinet cooling system 79,1
7 Servers, storage, cabinet cooling system 74,2
8 Servers, cabinet cooling system 75,0
9 Servers, cabinet cooling system 74,7
10 Servers, cabinet cooling system 73,9
11 Servers, cabinet cooling system 70,9
12 Servers, cabinet cooling system 70,9
13 Routers 75,2
14 Routers 80,3
15 Fire central (alarm and dischargers) 70,7
16 Console room 57,8
17 Archive 55,6
18 Operators room 58,5
19 Entrance hall 58,3
and IT equipment cooling fans. These fans operate at
discrete frequencies, and certain frequencies like blade
passing frequency (BPF), have more power than others.
BPF is the product of the fan rotation speed and the
number of fan blades, (4), [2]:
60
BR NN
BPF  (4)
where BPF is the basic frequency of fan tonal
components, NR is the fan rotation speed (rotations per
minute, RPM) and NB is the number of fan blades.
Numerous harmonics are present as multiples of BPF.
There is also a considerable noise in a frequency range of
2-7 kHz due to the turbulent air flow produced by the
fans. Above 7 kHz noise rapidly diminishes.
Figure 7. Spectrum of data center noise
D. Daily and Weekly Variations
Data centers have its temperature fluctuations that
may indirectly (due to different cooling requirements)
influence noise levels. This includes:
 daily variation caused by daily swings in the IT load
and outdoor temperature
 day-to-day variation caused by the weather
 IT load reduction on weekends
Daily noise variations measured at position 3 in data
center schematic from Fig. 6, are shown in Fig. 8.

MIPRO 2016/CTS 1355
Figure 8. Daily noise variations (linear amplitude scale, working day)
Noise level is quite constant with occasional spikes
due to human influence (opening data center doors,
opening server cabinets, loud communication etc.). Noise
levels averaged in four hour periods are shown in Table
V. Summary of daily variations, but averaged over 5s
period are given in Table VI. Weekly noise variations of
daily averages are given in Table VII.
TABLE V. DAILY NOISE VARIATIONS
Noise level (4
hour average) dBA
0-4 4-8 8-12 12-16 16-20 20-24
79,4 79,4 79,3 79,3 79,4 79,4
TABLE VI. DAILY NOISE VARIATIONS – SUMMARY
Noise level
(5s average) dBA
min max aver
78.5 88.3 79.4
TABLE VII. WEEKLY NOISE VARIATIONS – SUMMARY
Noise (daily
average) dBA
Sun Mon Thu Wen Thu Fri Sat
79,3 79,3 79,4 79,5 79,4 79,4 79,4
Data center where the measurements have taken place
is working with low utilization of computing resources
and main contributor to temperature changes is the
weather alone. As can be seen from our particular case,
noise variation during a period of one day and week are
negligible. In some data centers temperature fluctuations
may be up to 9 ºC depending on the time of day, season
and the current weather. Much lower temperature
excursions (just few degrees) are common in most data
centers.
V. MEASURED NOISE LEVELS, SPEECH
INTELIGIBILLITY AND COMMUNICATION DISTANCE
Several noise metrics have evolved for assessing the
influence of noise on speech, [12].
A. Speech Interference Level (SIL)
Speech Interference Level is defined as the arithmetic
average of the sound pressure levels at 500, 1000, 2000
and 4000 Hz octave bands, (5), [12, 13].
4
400020001000500 pppp LLLL
SIL 
 (5)
A-weighted sound level Lpa correlate well with SIL for
most sounds in aviation, [2]. Acceptable results of SIL
values maybe derived from A-weighted noise levels by
using the following approximate expression, (6), [13]:
10 pA
LSIL (6)
For data center noise spectral content shown in Fig. 7 and
put into (5), more appropriate expression would be (7):
5,12

pA
LSIL (7)
B. Articulation Index (AI)
Articulation Index (AI) is the value, between 0 and
1.0, which describes the masking of speech by
background noise. An AI of 1 means that all speech can
be understood, and 0 that no speech can be understood.
An AI < 0,05 means very poor speech intelligibility, and
an AI > 0.80 good speech intelligibility. AI is found by
evaluating the signal to noise ratio in specific frequency
bands, [12], according to (8), [14].

30
12
1




i
n
i
iSNRI
AI (8)
where
AI is the Articulation Index
Ii is the frequency importance function for band i
SNR is the Signal-to-Noise Ratio for band i
AI can be calculated from the 1/3 octave band levels
between 200 Hz and 6300 Hz center frequencies, using
the AI calculator, shown in Fig. 9, [15].
Figure 9. AI Calculator
C. Speech Intelligibility Index (SII)
The Speech Intelligibility Index (SII), [12, 16, 17], is
a standardized objective measure which is correlated with
the intelligibility of speech under a variety of adverse
listening conditions. SII is a function of the long-term
average spectrum of the speech and noise signals and is
based on the Articulation Index. SII and the AI are not
the same thing, but are quite similar. The SII, like the AI,
is a quantification of the proportion of speech information
that is both audible and usable for a listener. The SII is
computed as a product of the frequency band importance
function Ii, and the band audibility function Ai, (9).



n
i
ii AISII
1
(9)
The value of the SII varies from 0 (completely
unintelligible) to 1 (perfect intelligibility). SSI can be
determined using the SII Calculator, Fig. 10, [17].
Speech intelligibility in a data center is determined
according to mentioned intelligibility measures. Values
for SIL were determined using (5) and scaling the noise
values. In a similar way values for AI and SII were
determined using AI and SII calculator (for SII calculator

1356 MIPRO 2016/CTS
Figure 10. SII Calculator
user specified male speech spectrum normalized to 70 dB
SPL was supplied). Speech intelligibility values for
various noise levels are listed in Table VIII. At levels
within a data center, staff have to talk loudly to be heard.
D. Communication Distance
The distance between the talker and listener (i.e.
communication distance) is important when the
conversation takes place. Speech levels are reduced
typically by 6 dB for each distance doubling between the
talker and listener. Communication distances for various
values of SIL are shown in Fig. 11, [18], and for various
noise levels encountered in a data center in Table VIII.
Figure 11. Communication distance for various values of SIL
(adopted from [18])
TABLE VIII NOISE LEVELS, SPEECH INTELLIGIBILITY MEASURES AND
COMMUNICATION DISTANCES
Noise
level, dBA
Speech intelligibility1 Communication distance (m)
SIL AI SII normal speech loud speech
552 42,5 0,82 0,66 2,6 9,5
602 47,5 0,66 0,53 1,2 4,5
70 57,5 0,34 0,23 0,44 1,6
72 59,5 0,28 0,17 0,35 1,2
74 61,5 0,22 0,12 0,29 1,0
76 63,5 0,17 0,08 0,23 0,76
78 65,5 0,11 0,06 0,18 0,66
80 67,5 0,06 0,04 0,14 0,53
82 69,5 0,03 0,03 0,13 0,44
84 71,5 0,01 0,02 0,10 0,34
1normal speech, 2noise levels at accompanying spaces
VI. POSSIBLE NOISE REDUCTION SOLUTIONS
A. Passive Noise Reduction
Data center with its walls, floors and ceiling surfaces
provide little acoustic absorption. Conventional passive
acoustic absorption materials are not used because they
are highly flammable and may emit particulates that may
trigger fire detection sensors, [19]. Materials used in a
data center must be properly fire rated and fiber free.
They can be mounted on walls and suspended from
ceilings preventing reflections, reducing reverberation
and preventing noise build up. It is also possible to use
sound dampening server cabinets that use acoustic foam.
Passive noise protection headphones may be used for
tasks where no communication is needed. Conventional
hearing protection or noise protection headphones restrict
the workers ability to communicate. If communication
among workers is needed, headphones with integrated
wireless communication (like Bluetooth) could be used.
B. Active Noise Control
Noise within a data center consists of tonal and
broadband components. Tonal components originate from
HVAC fans and equipment cooling fans with pronounced
discrete spectrum. Broadband component originates from
turbulent airflow. Active Noise Control (ANC) is quite
successful at eliminating tonal components. Eliminating
these low to mid frequency components would greatly
reduce the amount of noise present in the data center.
ANC is generally not suitable for noise attenuation in
large spaces, but is mostly restricted to generating local
zones of silence. ANC over an extended region is not
possible in the data center room in general, [20]. This is
due to the random nature of the noise and the high modal
density of the room itself. If trying to suppress noise
within a large space better approach is to attenuate noise
at the source itself, like at ends of ventilation ducts.
ANC can be successfully applied for silencing HVAC
ventilation ducts, [21, 22], Fig. 12. Ventilation ducts
contain low to mid frequency tonal and broadband noise.
ANC systems for silencing ventilation ducts have been
developed decades ago, but with reduced cost of signal
processing hardware are becoming feasible for common
use. In laboratory, ANC can achieve 15-25 dB additional
noise reduction at frequency range of 50-350 Hz, [21],
and values of 10-15 dB are reported in a real world, [22].
Specially designed add-on unit can be installed at the end
of a pipe emitting or sucking air and generating noise.
ANC within servers, integrated with cooling fans, is
something to be common in a future. Such system uses a
magnetic field to generate minute vibrations in fan
blades. A microphone samples the noise and then tunes
the fan blade vibration to cancel it out, [23].
Figure 12. ANC in a ventilation duct

MIPRO 2016/CTS 1357
ANC headphones, [24], can provide greater deal of
protection, however most of ANC headphone designs are
optimized for aircraft use. Although ANC headphones are
successful at reducing noise, they also reduce the ability
to hear speech and remain aware of surroundings. Some
communication headphones combine speech enhancement
with noise suppression technology that elevates speech
and suppresses dangerous background noise so workers
can hear speech and stay protected in high noise
environments, [25]. Sometimes even situational awareness
with speech and noise direction recognition is preserved.
C. Lowering Cooling Requirements
HVAC is the main source of noise in a data center.
According to [6], 20% increase in cooling air speed
equates to a 4 dB increase in noise level. By reducing
HVAC activity it is possible to reduce noise.
Lower cooling requirements on HVAC can be
achieved by implementing higher degree of virtualization
and hence achieving higher CPU usage and the lower
number of necessary physical servers. Overall power
consumption, cooling requirements and noise levels go
down. However efficiency of power and cooling
infrastructure gets worse because of underloading, [26].
Another approach is raising target temperature within
a data center, [27]. The higher the target temperature in a
data center, the lower is the cooling requirement. HVAC
unit should not necessary run at full capacity at all times.
Increasing target temperature for just few degrees during
maintenance activities would lead to a quieter work
environment for employees. Recommendations for air
temperatures in data centers (Class 1 requirements) put
them in a range 18-27 °C (allowable 15-32 °C), [28]. As
an approximation it may be expected that 2°C increase in
data center temperatures would cause noise decrease of
3-5 dB, [6]. Server fans will, however, try to compensate
for diminished HVAC operation using the server built-in
temperature-based fan speed control, adding some noise.
VII. CONCLUSION
Measured noise levels within a data center were in
70-80 dBA range (as in other data centers). These levels
are below maximal acceptable eight hour exposure limit
but are close enough and it may be wise to use some kind
of hearing protection like noise protection headphones.
Prolonged exposure to such noise levels could produce
extra-aural health effects. High noise levels interfere with
performing cognitive tasks, influence autonomous system
and after the long-term exposure may cause behavior
modifications. Noise in a data center significantly
interferes with speech intelligibility (as reflected in
speech intelligibility measures), making communication
among personnel difficult (raising voice and reducing
communication distance). Lightweight ANC headphones
with build in wireless interphones would be beneficial.
Staff should perform as much as possible tasks remotely
from the office, using iLO console or RD connections.
Passive noise reduction with certified fire resistant
materials may be applied to data center surfaces. ANC is
suitable for HVAC ventilation ducts. During extensive
maintenance activities temperature throttling can be used.
REFERENCES
[1] TechMikeNY, Server Guide, Internet: https://techmikeny.com/
server-guide/ [February 29, 2016]
[2] Fan Noise Prediction, Internet: http://www.mne.psu.edu/
lamancusa/me458/11_fan.pdf [February 29, 2016]
[3] Noise Exposure - Permissible Level and Duration, Internet:
http://www.engineeringtoolbox.com/noise-exposure-level-duratio
n-d_717.html [February 29, 2016]
[4] Patricia T. Johnson, Noise Exposure: Explanation of OSHA and
NIOSH Safe‑Exposure Limits and the Importance of Noise
Dosimetry, ER078114 –B, Etymotic Research
[5] Sharp L. F., Swiney J. F., Dansby M. R., Hyatt S. C. & Schimmel
D. E.: Behavioral and physiological correlates of varying noise
environments, EPA-600/1-77-038, EPA, June 1977
[6] ASHRAE Publication, 2008 ASHRAE Environmental Guidelines
for Datacom Equipment, ASHRAE, 2008
[7] Pearsons K., Bennett Ricarda L. & Fidell S.: Speech levels in
various noise environments, EPA-600/1-77-025, EPA, May 1997
[8] P. H. Stypulkowski, “Circuitry options, signal processing under
altered signal conditions”, in Tobin, H (ed.): Practical Hearing Aid
Selection and Fitting, DIANE Publishing, 1997
[9] R. Jones, Factors That Affect Intelligibility in Sound Systems,
Meyer Sound, Internet: http://www.meyersound.com/support/
papers/speech/section2.htm [February 29, 2016]
[10] D. Alger, DC Deconstructed | Sounds of a Data Center, Cisco,
Internet: https://www.youtube.com/watch?v=EQTgpE35zSM&
nohtml5 [April 6 2015]
[11] National Standard for Occupational Noise [NOHSC: 1007(2000)],
2nd Ed., Canberra, Australia, July 2000
[12] Newman J. S. & Beattie K. R., Aviation Noise Effects, Report No.
FAA-EE-85-2, US DoT FAA, March 1985
[13] D. Miljković, J. Ivošević and T. Bucak, “Psycho-acoustical
ergonomics in a light aircraft interior”, Proc. of Ergonomics 2013,
pp. 111-116, 12-15 June 2013, Zadar, Croatia
[14] H. Bächler and A. Vonlanthen, “Signal processing tools to support
your manner of communication”, Phonak Focus 17, 1994
[15] Articulation Index, Internet: http://www.diracdelta.co.uk/science/
source/a/r/articulation%20index/source.html [February 29, 2016]
[16] American National Standard ANSI S3.5-1997, Methods for
Calculation of the Speech Intelligibility Index
[17] SII CALCULATION 1.0, A Software for ANSI S3.5-1997
[18] Transportation Safety Board of Canada (TSB), Railway
Investigation Report R98V0148, [February 29, 2016]
[19] Y. Joshi and P. Kumar (ed.), Energy Efficient Thermal
Management of Data Centers, Springer, 2012
[20] J. Daily, J. Esplin, Z. Collins and M. Shaw, Acoustical Analysis of
Active Control in the Server Room of a C7 Data Centers Colocation
Facility Feasibility Report, Brigham Young Univ., June 21, 2010
[21] M. Larsson, Active Noise Control in Ventilation Systems,
Blekinge Institute of Technology Licentiate Dissertation Series No
2008:12, Karlskrona, Sweden, 2008
[22] L. J. Gelin, “Active noise control: a tutorial for HVAC designers”,
ASHRAE Journal, August 1997
[23] G. Gasior, Noise-canceling Noctua CPU Cooler Coming in 2014,
The TechReport, Internet: http://techreport.com/news/24947/noise
-canceling-noctua-cpu-cooler-coming-in-2014 [February 29, 2016]
[24] C. Moy, Active Noise Reduction Headphone System, Internet:
http://cero-ruido.blogspot.com/2014/11/active-noise-reduction-hea
dphone-systems.html [February 29, 2016]
[25] Sensear, Data Center Industry Case Study, 2014
[26] Suzanne Niles, “Virtualization: ooptimized power and cooling to
maximize benefits”, White Paper #118, Schneider Electric, 2011
[27] To Lower the Noise Levels in Data Centers, Raise the
Temperature, Internet: http://www.itwatchdogs.com/environment
al-monitoring-news/data-center/to-lower-the-noise-levels-in-data-
centers,-raise-the-temperature-617409 [February 29, 2016]
[28] 2011 Thermal Guidelines for Data Processing Environments -
Expanded Data Center Classes and Usage Guidance, ASHRAE
TC 9.9, ASHRAE, 2011