Red Hat Enterprise Linux: The web performance leader
Upcoming SlideShare
Loading in...5
×
 

Red Hat Enterprise Linux: The web performance leader

on

  • 2,150 views

This reports illustrates how Red Hat Enterprise Linux beats Microsoft Windows as the best platform for enterprise web applications. Superior performance, coupled with significant cost savings derived ...

This reports illustrates how Red Hat Enterprise Linux beats Microsoft Windows as the best platform for enterprise web applications. Superior performance, coupled with significant cost savings derived from the ability to leverage the latest multicore x86 hardware as well as virtualization, adds up to competitive advantage.

Statistics

Views

Total Views
2,150
Slideshare-icon Views on SlideShare
2,150
Embed Views
0

Actions

Likes
0
Downloads
25
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Red Hat Enterprise Linux: The web performance leader Red Hat Enterprise Linux: The web performance leader Document Transcript

    • Red Hat enteRpRise Linux: The Clear leader for enTerprise Web appliCaTions Industry standard benchmarks illustrate that when you • Java Enterprise Edition (EE) application servers for need performance, scalability, and reliability for your web enterprise portals, service oriented architecture (SOA), applications, Red Hat Enterprise Linux should be your and business process management (BPM)6 first choice. • Large-scale SAP applications on x86_64 hardware7 As the fastest-growing server operating system in the world1, Red Hat Enterprise Linux is being deployed across • Large guests on virtualized x86_64 hardware8 a broad range of applications. Your dual Challenge: simulTaneouslY aChieving The reason is obvious when you consider the increasing CosT savings and high performanCe dependency on web technologies in all facets of company operations. Everything from customer sales to Human Amidst today’s turbulent economic conditions, the value Resources is being moved to the web browser as a client. proposition of getting the most from your IT environment However, as the end-client becomes mobile and easier to resonates more than ever. You face two key challenges manage, the burden on the server architecture increases. today when deploying enterprise applications: finding ways Whether it is a web server, an application server, or a data- to continually enhance performance and cutting costs. base, the superior performance and scalability of Red Hat Red Hat delivers on both. Enterprise Linux has been demonstrated in physical and First, performance. Both internal and external users expect virtual deployments. rapid response times. For internal users, productivity suf- Therefore, best practices for performance and scalability fers if your web applications are slow to respond. And when include deploying Red Hat Enterprise Linux throughout applications are externally focused — to be used by custom- your enterprise web applications. ers or partners — poor performance can drive users away, often to competitors. So, first and foremost, your web Independent tests performed under the guidelines imposed applications must be fast. by industry-standard benchmark standards2 found that Red Hat Enterprise Linux is the leading platform for run- And as you add users, performance can’t suffer: enterprise ning the following workloads and applications under a web applications must be able to accommodate hundreds three-tiered web application architecture: of thousands of users without performance degradation. After all, people expect a responsive, well-designed, and • Web servers 3 highly functional webpage. So your web applications must scale. • Java applications on x86_64 hardware4 But you also need to contain costs. This means taking • Large-scale online transaction processing (OLTP) advantage of the latest multi-core hardware as well as and database applications on x86_64 hardware5 leveraging virtualization for even greater cost savings. 1 Analysis based upon IDC Doc #218938/June 2009 2 TPC, SPEC, and SAP enforce strict policies on publishing benchmark results for competitive comparison purposes. All Red Hat reference architecture documents comply with these policies. 3 SPECweb2005 score = 71,045 6 SPECjAppServer2004 = 22,634 JOPS 4 SPECjbb2005 = 2,150,260 BOPS. 7 5,156 - 2 Tier SAP SD users 5 1,200,000 tpmC, $1.99/tpmC. 8 85 percent virtualization efficiency for 24vCPU guests running SAP SD www.redhat.com
    • In the past, you might have hesitated to deploy Linux — transactions of different types and complexity. The under- unsure if it was powerful or scalable enough to support lying database contained nine types of tables of varying your enterprise web applications. Due to the depth and record and data population sizes. Because TPC-C is an breadth of Red Hat's relationship with industry leaders, industry-standard benchmark, we can compare our results the proof has emerged and Red Hat Enterprise Linux has to those on Windows and legacy UNIX platforms.9 presented itself as a mature, mission-critical platform. As more IT managers are discovering, Red Hat Red Hat Enterprise Linux surpassed the Windows Server Enterprise Linux offers significant price/performance results in both overall throughput and in price/perfor- advantages over Microsoft Windows as well as other mance. The Red Hat Enterprise Linux and IBM results legacy operating systems. represented the first time that a technology combination exceeded one million transactions per minute (tpmC) on Over the past year, a number of leading technology firms — an x86-based system. Red Hat Enterprise Linux achieved including IBM, HP, and SAP — invited Red Hat to participate 1.2 million tpmC at $1.99 on an 8 processor x86_64 server. in industry-standard benchmarks that compared Red Hat The nearest Windows Server result on similar hardware Enterprise Linux to Microsoft Windows and other operat- was .84 million tpmC at $3.46 (see Figure 1 and Figure 2). ing systems under scenarios comparable to those you deal with every day. The benchmarks show Red Hat Conclusion: From both a performance and a cost perspec- Enterprise Linux is the leader in web-based performance. tive, Red Hat Enterprise Linux is the superior platform for OLTP on x86_64 systems when compared to Windows. red haT enTerprise linux aT The daTabase Tier besT TpC-C priCe/ performanCe for 8 proCessor Databases are the heart of your secure, scalable, and x86_64 servers (deC 2008) reliable web application. As you add users and increase 4.00 application complexity, you must be careful that perfor- 3.50 3.46 mance doesn’t suffer — both on bare metal and in virtual- 3.00 ized environments. $ / Ktpmc 2.50 2.00 red haT enTerprise linux proven besT plaTform 1.99 1.50 for online TransaCTion proCessing 1.00 As the class of systems that manages mission-critical 0.50 online applications for enterprises in industries ranging 0.00 from banking, to manufacturing, to retail, online trans- action processing (OLTP) is increasingly performed on Red Hat Enterprise Linux Win Server 2003 clusters of x86 machines. OLTP can refer to transactions Benchmarked Systems that are either executed online, as in ATM transactions, or placed in a queue to be executed later, as in many order Figure 1: Red Hat Enterprise Linux had the best OLTP processing systems. system price/performance when compared In the x86 world, enterprises have the choice of x86 and to Windows. Windows platforms for their OLTP applications. Recently, IBM and Red Hat ran the TPC Benchmark™C (TPC-C) on 9 The TPC-C Benchmark was performed with DB2 on IBM Systems x3960 IBM's latest x86 systems, testing a mix of five concurrent M2 using Red Hat Enterprise Linux operating at more than one million transactions per minute (tpmC). 2 www.redhat.com
    • besT TpC-C ThroughpuT for x86_64 oraCle olTp performanCe sCaling on red haT servers (oCT 2008) enTerprise linux 1,400,000 350 1,200,000 300 1,200,000 Ktpmc Thousands of TPM Harpertown 1,000,000 250 Tigerton 800,000 841,800 200 Dunnington 600,000 150 400,000 100 200,000 50 0 Red Hat Enterprise Linux Win Server 2003 0 5 10 15 20 25 30 Benchmarked Systems # Cores Figure 2: Red Hat Enterprise Linux delivered the best throughput when compared to Windows. Figure 3 red haT enTerprise linux sCales To handle mission-CriTiCal oraCle 10g daTabases Driven by cost savings, more and more enterprises are migrating applications from UNIX on proprietary RISC hardware to Red Hat Enterprise Linux running on commod- ity x86 machines. To manage the demands of their largest mission-critical databases, such businesses need the option to scale up the number of processors in each system. Red Hat recently ran benchmarks that tested the verti- cal scaling for Oracle database performance on Red Hat Enterprise Linux on servers with increasingly larger num- bers of x86_64 cores.10 Previous benchmarks demon- strated excellent scalability up to eight cores. But in March 2009, Red Hat performed a test dramatizing that Oracle OLTP applications could scale vertically up to 24 cores (see Figure 3). 10 The scaling of Oracle 10g as tested using Red Hat Enterprise Linux 5 on Intel Xeon-based servers www.redhat.com 3
    • oraCle 10g performanCe on red haT enTerprise linux 5 virTualizaTion sCales sap aChieves Top virTualizaTion effiCienCY boTh horizonTallY and verTiCallY WiTh red haT enTerprise linux virTualizaTion No matter whether an enterprise wants to scale It is the common wisdom that server efficiency can its Oracle 10g databases vertically or horizontally, drop — often precipitously — in virtualized environments. Red Hat Enterprise Linux with integrated virtualiza- This has discouraged many enterprises from deploying tion allows the customer to achieve its performance virtualization for their largest and most mission-critical goals. Almost every virtualization solution supports database applications. But in November 2008, when increased utilization of your system by adding mul- Red Hat tested the performance hit that a very large- tiple smaller guests. Red Hat's virtualization tech- scale SAP SD application took in a virtualized environ- nology also allows you to scale up and add up to 32 ment under Red Hat Enterprise Linux,11 it proved to be virtual CPUs to a single guest. As internal bench- minimal. With an efficiency rate of 85 percent, the virtu- marking shows, scaling up virtual CPU systems pro- alized SAP application was able to accommodate 4,400 vides you the same benefits that you would expect users when compared to the 5,156 users supported on from adding physical CPUs. (Figure 4). This allows bare metal (see Figure 6). them to accommodate ever-larger mission-critical workloads. Moreover, because of the higher virtual CPU limit per guest made possible by Red Hat Enterprise Linux 5 virtualization, Red Hat Enterprise Linux outperformed relaTive performanCe of daTabase Workload in a virTual guesT other leading virtualization solutions that were also tested at SAP's Linux Labs (see Figure 5). 11 10 Comparison of bare-meTal versus para-virTualized 9 guesT performanCe using sap-sd benChmark on red 8 haT enTerprise linux 5.2 Realtive Performance 7 6 6,000 85% 5 5,000 5,156 SAP-SD Users 4 4,000 841,800 3 3,000 2 2,000 1 1,000 0 1 8 16 24 Core Bare Metal 24 vCPU PV Guest # cores/ guest Figure 4: Enterprises can use Red Hat Enterprise Figure 5: Red Hat Enterprise Linux virtualization is the Linux with Oracle 10g to scale up by assigning more best platform for running large web applications such CPUs to a single guest. as SAP as guests on x86_64 hardware. 11 Under the SAP SD Benchmark using DB2 running Red Hat Enterprise Linux 5 Virtualization on IBM System x3850 M2 4 www.redhat.com
    • The business logiC Tier besT speCJbb2005 bops for x86_64 servers (mar 2009) Users simply won’t tolerate a slow-moving web application that performs more sluggishly than the other applications 2,500,000 SpecJBB2008 BOPS they use. And you have to continue keeping your eye on 2,000,000 2,150,260 costs. After all, the lower the throughput, the more hard- 1,500,000 ware you need to run your application. That adds up. 1,000,000 1,181,782 1,037,851 Further benchmarks that compared Red Hat 500,000 659,034 Enterprise Linux to other operating environments found 0 that Red Hat Enterprise Linux achieves more business Red Hat Win Server Solaris Win Server processes per minute and enables more processing with Enterprise Linux 2008 2003 less hardware. Benchmarked Systems red haT enTerprise linux beaTs WindoWs and solaris running Java on x86_64 hardWare Figure 6: Red Hat Enterprise Linux was by far the Enterprises need their business-critical Java applictions — best-performing operating system for running Java both internal and customer-facing — to perform at top applications. speed or risk having employee productivity and customer satisfaction deteriorate. In February 2009, Red Hat simu- lated an order processing application scenario for a whole- red haT enTerprise linux is The besT operaTing sale supplier to test the performance capabilities of various sYsTem for Web servers operating systems running on Intel Xeon-based hardware.12 Red Hat Enterprise Linux beat both Windows and Solaris In the online world, responsiveness is everything. It is the in benchmarks that measured business operations per responsibility of the web server to manage the connections second (BOPS) as well as the BOPS per each Java virtual and deliver the content. Not only is this an I/O-intensive machine (JVM) instance (see Figure 6). function, but in secure SSL environments, it is also com- putationally intensive. Even a few seconds delay can lose customers, slow down worker productivity, and hurt the bottom line. Small wonder that businesses are constantly seeking the most robust and high-performing operating system for the web servers running their mission-critical online applications. After testing the performance of Red Hat Enterprise Linux using three common enterprise workloads — banking, ecom- merce, and support — Red Hat Enterprise Linux emerged the clear winner.13 13 The tests were conducted in accordance with SPECweb2005, the next- 12 Test was run using the SPECjbb 2005 Benchmark using Red Hat Enterprise generation SPEC benchmark for evaluating the performance of World Wide Linux 5.3 on an 96-core Intel Xeon-based NEC server. Web servers. www.redhat.com 5
    • Red Hat Enterprise Linux achieved the highest SPECweb2005 score on benchmarks completed in besT speCWeb2005 (4-soCkeT x86_64 servers) January 2009. Of the top 20 published SPECweb2005 results, 19 were achieved using Red Hat Enterprise Linux. 70,000 There are numerous results for SpecWeb2005 that docu- SPECweb2005 Results 60,000 62,098 ment the increase in performance from generation to 50,000 generation. However, none of them have been run on a 40,000 Windows Server platform. 39,793 30,000 20,000 besT speCWeb2005 (2-soCkeT x86_64 servers) 10,000 0 80,000 SunFire HP DL580 G5 70,000 X4450 - SuSE Red Hat Enterprise Linux 71,045 SPECweb2005 Results 60,000 Benchmarked Systems 50,000 Figure 8: Red Hat Enterprise Linux achieved the best 40,000 results on web server benchmarks conducted on 30,000 four-socket x86_64 machines. 20,000 22,332 10,000 0 red haT enTerprise linux besT plaTform PR4700D HP DL380 G6 (Xeon 5570) - for running Java enTerprise ediTion (Xeon 5365) - SuSE Red Hat Enterprise Linux appliCaTion servers Benchmarked Systems Whether needing a platform to support their SOA or busi- Figure 7: Red Hat Enterprise Linux achieved the best ness process management (BPM) efforts, enterprises score on two-socket x86_64 hardware. require a robust, scalable, and high-performing plat- form for Java Enterprise Edition (EE) application serv- ers. Through benchmarks performed in February 2009, Red Hat Enterprise Linux beat out HP-UX, AIX, and Solaris by comparing jAppServer operations per second (JOPS) achieved.14 (See Figure 9.) 14 SPECjAppServer2004 Benchmark using Red Hat Enterprise Linux 5.2 on an IBM BladeCenter HS21. 6 www.redhat.com
    • In November 2008, Red Hat Enterprise Linux achieved the besT speCJappserver2004 best 24-core performance to date — beating both Solaris and Windows — on x86_64 servers by scaling to support 25,000 simultaneous transactions by 5,156 SAP SD users (see SPECjAppServer2004 JOPS 22,634 Figure 10). 20,000 21,602 15,000 14,004 10,000 2-Tier sap sd benChmark resulTs (nov 2008) 9,501 5,000 5,200 0 5,155 5,156 5,100 5,135 Red Hat HP-UX/ AIX/ Solaris/ Enterprise Linux/ Itanium Power SPARC x86_64 5,000 Benchmarked Systems Certified SAP SD Users 4,900 4,800 Figure 9: Red Hat Enterprise Linux beat out HP-UX, AIX, and Solaris when benchmarking the perfor- 4,700 mance of Java EE application servers. 4,600 4,600 4,500 4,400 puTTing iT all TogeTher: red haT enTerprise 4,300 linux sCales beTTer Than WindoWs for large SunFire X4450 - Fujitsu PRIMERGY HP Proliant IBM x3850M2 - sap appliCaTions Solaris - MaxDB RX600S4 - Win2K3 - DL580G5 - Win2K3 - Red Hat Enterprise Linux 5 -DB2 SQLServer SQLServer To determine whether Red Hat Enterprise Linux can scale Benchmarked Systems (Intel Xeon X7460 4x6=24 Core Based Servers) sufficiently to support the large number of users of the SAP Sales and Distribution (SD) applications that global Figure 10: Red Hat Enterprise Linux outperformed retailers depend on to run their businesses, Red Hat ran Windows and Solaris on SAP SD benchmark in the SAP SD Benchmark.15 This benchmark has become a November 2008. de facto standard for evaluating the performance of many ERP solutions, and tests a scenario in which a customer places an order for five separate products, and encom- passes all processes and processing involved in moving those products from inventory through delivery and cus- tomer invoicing. Conventional wisdom in the marketplace has been that Red Hat Enterprise Linux cannot scale to accommodate a large number of customers under a scenario of this sort. This test proved otherwise. 15 The benchmark used DB2 and Red Hat Enterprise Linux 5 on IBM System x3850 M2 www.redhat.com 7
    • ConClusion: no Compromises WiTh red haT enTerprise linux These rigorous tests, run in conformance with accepted industry-standard benchmarks, illustrate how Red Hat Enterprise Linux beats Microsoft Windows as the best plat- form for enterprise web applications. This superior perfor- mance, coupled with significant cost savings derived from the ability to leverage the latest multicore x86 hardware as well as virtualization, adds up to competitive advantage for you. Whether it's migration, consolidation, expansion, or just plain good architectural capacity planning, Red Hat Enterprise Linux is your best choice for an enterprise oper- ating system that will enable all of those efforts. red haT sales and inquiries NORTH AMERICA 1–888–REDHAT1 www.redhat.com Copyright © 2009 Red Hat, Inc. Red Hat, Red Hat Enterprise Linux, the Shadowman logo, JBoss, MetaMatrix, and RHCE are trademarks of www.redhat.com Red Hat, Inc., registered in the U.S. and other countries. Linux® is the registered trademark of Linus Torvalds in the U.S. and other countries. #1201160_0809