MAY 2013

A PRINCIPLED TECHNOLOGIES SUMMARY

Commissioned by Red Hat, Inc.
SERVER PERFORMANCE WITH RED HAT ENTERPRISE LINUX 6 VS.
MICROSOFT WINDOWS SERVER 2012

Selecting the right operating system for your datacenter can potentially
maximize the performance levels you get and improve end-user experience. When
faced with a choice between Microsoft Windows Server 2012 and Red Hat Enterprise
Linux 6, it is important to assess the capabilities of these operating systems on your
hardware with both out-of-box and tuned configurations to ensure you can get the most
from your servers.
In the Principled Technologies labs, we investigated the server performance on
hardware running Red Hat Enterprise Linux 6 and Microsoft Windows Server 2012 by
testing several different subsystems, including CPU, memory, disk I/O, and network. In
nearly every test, we found that Red Hat Enterprise Linux 6 outperformed its Microsoft
competitor, and found that tuning with the Red Hat solution generally increased
performance even more.
If an operating system can get more out of each server subsystem, as Red Hat
Enterprise Linux 6 did in our tests, it follows that it has the potential to improve
application performance across the board. Here, we present a brief summary of our
findings. For more detail, check out our full reports, which we link below.





A Principled Technologies summary 2


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Testing Java application performance
A Java Virtual Machine (JVM) provides a sufficiently rich abstraction layer to run
applications independent of particular computer hardware implementation. Testing
Java performance with the SPECjbb2013
1
benchmark can give an indication of the
Java application performance a solution provides. As our results indicate, Red Hat
Enterprise Linux 6 with OpenJDK
2
outperformed Microsoft Windows Server 2012 with
Java HotSpot
3
on the industry-standard SPECjbb2013 benchmark on both of the
reported metrics, max-jOPS and critical-jOPS, using small heap size. With large heap
size, the Red Hat/OpenJDK solution delivered 34,129 max-jOPS and 22,126 critical-jOPS,
the best reported critical operations score as of June 30, 2013, while the Microsoft/Java
HotSpot solution could not produce a qualifying benchmark result.
4
Figure 1 shows the
small heap scores for both systems and the large heap score for Red Hat Enterprise
Linux 6 with OpenJDK.
Figure 1: SPECjbb2013 scores
for Red Hat Enterprise Linux 6
with OpenJDK and Microsoft
Windows Server 2012 with
Java HotSpot. Higher numbers
are better.
22,126
34,129
14,655
34,951
14,414
33,718
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
critical-jOPS max-jOPS
S
P
E
C
j
b
b
2
0
1
3

j
O
P
S
SPECjbb2013 scores
Red Hat Enterprise Linux 6 with OpenJDK (large heap)
Red Hat Enterprise Linux 6 with OpenJDK (small heap)
Microsoft Windows Server 2012 with Java HotSpot (small heap)

For details on our SPECjbb2013 testing, read the full report at
www.principledtechnologies.com/RedHat/RHEL6_jbb_0613.pdf.
Testing file system I/O performance
When choosing an operating system platform for your servers, you should know
what I/O performance to expect from the operating system and file systems you select.


1
SPEC and SPECjbb are trademarks of the Standard Performance Evaluation Corp. (SPEC). See www.spec.org for more information.
2
OpenJDK is a trademark of Oracle, Inc.
3
Java and Java HotSpot are trademarks of Oracle, Inc.
4
Small heap (20GB) OpenJDK and Java HotSpot JVM results were not submitted to SPEC for publication. Large heap OpenJDK results
are official SPECjbb2013 results at spec.org. See www.spec.org/jbb2013/results/res2013q2/jbb2013-20130319-00013.html for more
details on large heap results.




A Principled Technologies summary 3


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Using the IOzone Filesystem Benchmark in our tests, we found I/O performance of file
systems on Red Hat Enterprise Linux 6 was better than the file systems available on
Microsoft Windows Server 2012, with both out-of-the-box and optimized
configurations. Using default native file systems, ext4 and NTFS, we found that Red Hat
Enterprise Linux 6 outperformed Windows Server 2012 by as much as 65.2 percent out-
of-the-box, and as much as 33.4 percent using optimized configurations. Using more
advanced native file systems, XFS and ReFS, we found that Red Hat Enterprise Linux 6
outperformed Windows Server 2012 by as much as 31.9 percent out-of-the-box, and as
much as 48.4 percent using optimized configurations. Figure 2 shows file system
performance using the in cache test case our full test included direct I/O and out-of-
cache.
Figure 2: Comparison of the
I/O performance in KB/s for
the four file systems using the
in-cache method. These
throughputs are the geometric
average of the 13 IOzone tests.
Higher throughput is better.
0
1,000,000
2,000,000
3,000,000
4,000,000
5,000,000
6,000,000
7,000,000
Optimized Out-of-box
A
v
e
r
a
g
e

p
e
r
f
o
r
m
a
n
c
e

(
K
B
/
s
)
Comparison of file system performance - In cache
Red Hat Enterprise Linux 6 ext4 Microsoft Windows Server 2012 NTFS
Red Hat Enterprise Linux 6 XFS Microsoft Windows Server 2012 ReFS

To learn more about our complete I/O testing, see the full report at
www.principledtechnologies.com/RedHat/RHEL6_IO_0613.pdf.
Testing the CPU with SPEC CPU2006 and LINPACK
In our CPU tests, we found that the open-source Red Hat Enterprise Linux 6
solution performed as well or better than Microsoft Windows Server 2012. In our SPEC
CPU2006 tests, the Red Hat Enterprise Linux 6 solution achieved consistently higher
scores than the Windows Server 2012 solution. Figures 3 and 4 show the scores that the
systems achieved on both parts of the benchmark: SPEC CINT2006 and SPEC CFP2006.




A Principled Technologies summary 4


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Figure 3: SPEC CPU2006
results, in SPEC CINT2006
scores, for the two solutions.
Higher numbers are better.
639 640
623 621 618 619
593 591
0
100
200
300
400
500
600
700
Out-of-box Optimized Out-of-box Optimized
SPECint_rate2006 SPECint_rate_base2006
S
P
E
C

C
I
N
T
2
0
0
6

s
c
o
r
e
SPEC CINT2006 scores
Red Hat Enterprise Linux 6 Microsoft Windows Server 2012


Figure 4: SPEC CPU2006
results, in SPEC CFP2006
scores, for the two solutions.
Higher numbers are better.
422 422
408
404 406 405
399 397
0
50
100
150
200
250
300
350
400
450
Out-of-box Optimized Out-of-box Optimized
SPECfp_rate2006 SPECfp_rate_base2006
S
P
E
C

C
F
P
2
0
0
6

s
c
o
r
e
SPEC CFP2006 scores
Red Hat Enterprise Linux 6 Microsoft Windows Server 2012


We also used the LINPACK benchmark to test the floating point performance of
the platforms out-of-box and optimized. As Figure 5 shows, Red Hat Enterprise Linux 6
outperformed Windows Server 2012 at high thread counts in our tests. In addition,
tuning Red Hat Enterprise Linux 6 increased performance steadily, while optimizing
Windows Server 2012 had less of an effect on its performance.




A Principled Technologies summary 5


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Figure 5: LINPACK floating
point performance results for
the two operating system
solutions.
0
50
100
150
200
250
300
1 2 4 8 16
P
e
r
f
o
r
m
a
n
c
e

(
G
F
l
o
p
s
)
Number of threads
LINPACK: Average floating point performance
Red Hat Enterprise Linux 6 optimized Microsoft Windows Server 2012 optimized
Red Hat Enterprise Linux 6 out-of-box Microsoft Windows Server 2012 out-of-box


For complete details about our SPEC CPU2006 and LINPACK testing, read the full
report at www.principledtechnologies.com/RedHat/RHEL6_CPU_RAM_0613.pdf.
Testing memory performance
Random access memory (RAM) is one of the most vital subsystems that can
affect the performance of business applications. We used the STREAM benchmark to
measure the memory bandwidth available with both Red Hat Enterprise Linux 6 and
Microsoft Windows Server 2012. In our memory bandwidth tests, the Red Hat
Enterprise Linux 6 solution outperformed the Windows Server 2012 solution at mid-
range thread counts. Figures 6 and 7 show the performance of Red Hat Enterprise Linux
6 and Microsoft Windows Server 2012, both out-of-box and optimized.




A Principled Technologies summary 6


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Figure 6: Out-of-box memory
bandwidth comparison using
the STREAM benchmark.
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
1 2 4 8 16
S
u
s
t
a
i
n
e
d

m
e
m
o
r
y

b
a
n
d
w
i
d
t
h

(
M
B
/
s
)
STREAM tests and number of cores
Memory bandwidth comparison:
Out-of-box configurations
Red Hat Enterprise Linux 6 Microsoft Windows Server 2012


Figure 7: Optimized memory
bandwidth comparison using
the STREAM benchmark
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
C
o
p
y
S
c
a
l
e
A
d
d
T
r
i
a
d
1 2 4 8 16
S
u
s
t
a
i
n
e
d

m
e
m
o
r
y

b
a
n
d
w
i
d
t
h

(
M
B
/
s
)
STREAM tests and number of cores
Memory bandwidth comparison:
Optimized configurations
Red Hat Enterprise Linux 6 Microsoft Windows Server 2012


For complete details about our RAM testing using STREAM, read the full report
at www.principledtechnologies.com/RedHat/RHEL6_CPU_RAM_0613.pdf.




A Principled Technologies summary 7


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
Testing networking performance
Because applications may not typically manage networking resources directly
and instead may rely on operating systems to do so, the operating system you select
may have a direct impact on TCP and UDP performance available to your applications
and users. Throughout our network tests, we found that the open-source Red Hat
Enterprise Linux 6 solution delivered up to three times better TCP throughput than
Microsoft Windows Server 2012 in an out-of-box configuration and up to two times
better throughput in an optimized configuration. In addition, Red Hat Enterprise Linux 6
delivered better UDP throughput at various message sizes. Figure 8 shows the
performance of both solutions in the two configurations.
Figure 8: TCP throughput, in
10
6
b/s, the two solutions
achieved using different size
messages. Higher numbers are
better.
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
T
h
r
o
u
g
h
p
u
t

(
1
0
6
b
/
s
)
Message size (bytes)
TCP throughput
Red Hat Enterprise 6 OOB Red Hat Enterprise 6 OPT
Windows Server 2012 OOB Windows Server 2012 OPT


To learn more about our network testing, read the full report at
www.principledtechnologies.com/RedHat/RHEL6_network_0613.pdf.
IN CONCLUSION
Throughout all of our subsystem tests, we found that the open-source Red Hat
Enterprise Linux 6 operating system generally performed better than Microsoft
Windows Server 2012. In general, we also found that tuning the operating system
allowed us to get even greater performance out of the system running Red Hat
Enterprise Linux 6.




A Principled Technologies summary 8


Server performance with Red Hat Enterprise Linux 6 vs.
Microsoft Windows Server 2012
ABOUT PRINCIPLED TECHNOLOGIES


Principled Technologies, Inc.
1007 Slater Road, Suite 300
Durham, NC, 27703
www.principledtechnologies.com
We provide industry-leading technology assessment and fact-based
marketing services. We bring to every assignment extensive experience
with and expertise in all aspects of technology testing and analysis, from
researching new technologies, to developing new methodologies, to
testing with existing and new tools.

When the assessment is complete, we know how to present the results to
a broad range of target audiences. We provide our clients with the
materials they need, from market-focused data to use in their own
collateral to custom sales aids, such as test reports, performance
assessments, and white papers. Every document reflects the results of
our trusted independent analysis.

We provide customized services that focus on our clients individual
requirements. Whether the technology involves hardware, software, Web
sites, or services, we offer the experience, expertise, and tools to help our
clients assess how it will fare against its competition, its performance, its
market readiness, and its quality and reliability.

Our founders, Mark L. Van Name and Bill Catchings, have worked
together in technology assessment for over 20 years. As journalists, they
published over a thousand articles on a wide array of technology subjects.
They created and led the Ziff-Davis Benchmark Operation, which
developed such industry-standard benchmarks as Ziff Davis Medias
Winstone and WebBench. They founded and led eTesting Labs, and after
the acquisition of that company by Lionbridge Technologies were the
head and CTO of VeriTest.
Principled Technologies is a registered trademark of Principled Technologies, Inc.
All other product names are the trademarks of their respective owners.
Disclaimer of Warranties; Limitation of Liability:
PRINCIPLED TECHNOLOGIES, INC. HAS MADE REASONABLE EFFORTS TO ENSURE THE ACCURACY AND VALIDITY OF ITS TESTING, HOWEVER,
PRINCIPLED TECHNOLOGIES, INC. SPECIFICALLY DISCLAIMS ANY WARRANTY, EXPRESSED OR IMPLIED, RELATING TO THE TEST RESULTS AND
ANALYSIS, THEIR ACCURACY, COMPLETENESS OR QUALITY, INCLUDING ANY IMPLIED WARRANTY OF FITNESS FOR ANY PARTICULAR PURPOSE.
ALL PERSONS OR ENTITIES RELYING ON THE RESULTS OF ANY TESTING DO SO AT THEIR OWN RISK, AND AGREE THAT PRINCIPLED
TECHNOLOGIES, INC., ITS EMPLOYEES AND ITS SUBCONTRACTORS SHALL HAVE NO LIABILITY WHATSOEVER FROM ANY CLAIM OF LOSS OR
DAMAGE ON ACCOUNT OF ANY ALLEGED ERROR OR DEFECT IN ANY TESTING PROCEDURE OR RESULT.

IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES, INC. BE LIABLE FOR INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES IN
CONNECTION WITH ITS TESTING, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES,
INC.S LIABILITY, INCLUDING FOR DIRECT DAMAGES, EXCEED THE AMOUNTS PAID IN CONNECTION WITH PRINCIPLED TECHNOLOGIES, INC.S
TESTING. CUSTOMERS SOLE AND EXCLUSIVE REMEDIES ARE AS SET FORTH HEREIN.