Dimension Data Public Compute-as-a-Service (CaaS)

697 views
612 views

Published on

Dimension Data’s cost effective, secure, reliable, and high performance cloud computing service has been put to the test. According to Tolly Group’s proven benchmarks report, Dimension Data can accelerate your clients’ adoption of public and private cloud. Dimension Data offers a network-centric cloud approach which provides clients with automated and self-service control of Cisco based computing and networking.

Below are some examples of how we compete with other cloud service providers, including Amazon Web Services, IBM Smartcloud, and Rackspace:

1. Proof point: Of all Dimension Data’s cloud services clients, 40% trust Dimension Data’s Cisco Powered platform to run production applications.

2. Proof point: Dimension Data client, Glassbeam, saves 50% on annual operating costs by migrating Glassbeam’s SaaS applications from an internal dedicated server environment onto Dimension Data’s virtual cloud platform.

3. Proof point: After evaluating a host of cloud service providers, HR benefits provider, ClearBenefits, chose Dimension Data to deliver 99.9% reliability for their SaaS application.

The report is also posted at
www.cisco.com/go/cloudforpartners located under Technology Partners.

Published in: Technology
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
697
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide

Dimension Data Public Compute-as-a-Service (CaaS)

  1. 1. Dimension Data Public Compute-as-a-Service (CaaS) Infrastructure-as-a-Service: Cloud Server & Network Performance vs. Amazon, IBM and Rackspace TEST HIGHLIGHTS 2 Delivered over twice the memory throughput as thenearestcompetitorand5XthatofAmazon 1 Completed the CPU-intensive test faster than AmazonandRackspaceinallcategories 4 Delivered true Gigabit Ethernet-class throughput - 6.5XthatofAmazonand9.6XthatofRackspace Delivered file I/O performance 3X to 6X that of the otherofferings 3 EXECUTIVE SUMMARY The emergence of cloud computing as a viable path for implementing enterprise-class computing solutions brings with it many opportunities. Cloud computing also challenges prospective customers to understand the actual performancedeliveredbyvarioussolutionproviders. Dimension Data commissioned Tolly to benchmark the system performance and networking throughput of web/app servers running on its public cloud solution and compare them to similar configurations running on platforms offeredbyAmazonWebServices,IBMandRackspace. Testing included benchmarking key system resources and network throughput across three categories of web/application cloud servers. The Dimension Data cloud servers showed consistently high performance across the range of resourcesbenchmarked. ...<continuedonnextpage> Source: Tolly, May 2013 © 2013 Tolly Enterprises, LLC Page 1 of 7Tolly.com #213131 July 2013 Commissioned by Dimension Data Linux Cloud Server CPU Performance C-Ray 1.1 Benchmark (as reported by Phoronix Test Suite 3.6.1) 0 200 400 600 800 1000 144 289 101 190 227 433 909 141 284 606 BenchmarkCompletionTime(seconds) Dimension Data Amazon Web Services IBM SmartCloud Rackspace Notes: For Amazon Web Services, the number shown is the number of EC2 units. Neither IBM nor Rackspace offers a 1 vCPU solution. IBM running RHEL 6.3, all others running Ubuntu 10.04 LTS Server. All systems were 64-bit. Small 1 vCPU Figure 1 Medium LargeSystem Category 2 vCPUs 4 vCPUs 909 DimensionDataCloudServers: Lower numbers are better
  2. 2. While there is no industry standard for sizing of servers, most servers are defined by the key resources of CPU and memory. These tests were run across three categories of systems defined by the following virtual CPU (vCPU) and system random access memory (RAM) configurations found in parentheses after each designation: Small (1/2), Medium (2/4), and Large (4/8). It was not always possible to match these requirements so the closest configurations were used and noted. Ubuntu 10.4 Server was the primary test platform. Since IBM does not offer that platform, Red Hat EL 6.3 was used for IBM. For Amazon, m1 instances (small, medium, large) were used. All Rackspace servers wereNextGeneration. The open-source Phoronix Test Suite (PTS) was used to benchmark CPU, memory and file capabilities. Iperf was used to benchmarknetworkthroughput. Test Results CPU The PTS C-Ray benchmark is a compute- intensive program and, simply put, a more powerful CPU will complete the benchmarkfaster. The Dimension Data servers completed the test faster than Amazon and Rackspace systems in all categories, second only to IBM. (IBM and Rackspace do not offer 1vCPU servers.) In fact, Amazon took 50% longer than Dimension Data to complete a series of compute-intensive tests. See Figure1. Memory The PTS RAMSpeed benchmark drives system memory operations. Dimension Data outperformed all other solutions across all categories. The results were most dramatic in the large system category with Dimension Data delivering over twice as great memory throughput as the nearest competitor and over 5 times the throughput of the Amazon solution. Dimension Data had significantly dramatic performance benefits over Amazon in the otherconfigurationstested.SeeFigure2. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 2 of 7Tolly.com Tested May 2013 Dimension Data Infrastructure- as-a-Service Cloud Server & Network Performance Dimension Data Amazon Web Services IBM SmartCloud Rackspace Source: Tolly, May 2013 Figure 2 0 5000 10000 15000 20000 7,818 6,522 8,772 9,985 3,200 2,523 1,225 18,542 10,831 3,110 MemoryOperationsPerSecond(Average) Linux Cloud Server System Memory Performance RAMSpeed 3.5 Benchmark (as reported by Phoronix Test Suite 3.6.1) 4 GB RAM 2 GB RAM 8 GB RAM Note: Neither IBM nor Rackspace offers a 1 vCPU solution. IBM running RHEL 6.3, all others running Ubuntu 10.04 LTS Server. All systems were 64-bit. Medium LargeSmallSystem Category 8,772 18,542
  3. 3. File I/O The PTS PostMark benchmark measured small file transaction performance on the local disk using the default file system for each solution. The Dimension Data cloud server file performance exceeded all other cloud providers by 3X to 6X. Dimension Data’s file I/O, as measured by transactions per second, of 3,472 was 2.7X that of Amazon and 5.4X that of Rackspace for a mediumsystem.SeeFigure3. NetworkThroughput This test used the open-source Iperf network benchmarking program to measure the bidirectional network throughputbetweenapairofcloudservers on the same internal data center network. Default configurations were used for each solution.Eachsystemundertestwaspaired with an Iperf partner system configured to match the large configuration. This maximized the throughput of the solution undertest. Where the various vendors were typically specific about the CPU and memory resource provided with each server category,thatwasnotusuallythecasewith the characteristics of the network interface. As the LAN capability is also virtualized, actual throughput can be much lower than the nominal “Gigabit Ethernet” (GbE) interfacethatistypical. Cloud vendors can limit the maximum throughput of a virtual network interface using readily available rate limiting functionality. In contrast, vendors with 10GbE backbone links can make that bandwidth available to the nominal GbE interface. The maximum throughput of a physical GbE port is 2Gbps of bidirectional traffic. The ports are full-duplex and a full stream of1Gbpstrafficcanflowineachdirection. For this test, Tolly engineers included the small instance from Rackspace even though it was a 2vCPU instance to determinewhatnetworkthroughputcould be expected from that category of Rackspace server. (The Amazon and Dimension Data instances were 1vCPU systems.) The results show that only Dimension Data delivers true Gigabit Ethernet-class throughput - providing greater than 1Gbps of throughput even in the small category. Dimension Data’s network throughput of 2.25Gbps is 6.5X that of Amazon and 9.6X thatofRackspace. In fact, across all server categories, Rackspace appears to apply severe rate limiting. The comparison with Dimension Dataisquitedramatic. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 3 of 7Tolly.com Source: Tolly, May 2013 Figure 3 0 1000 2000 3000 4000 659642 527 684 1,3421,278 402 3,4723,472 1,448 TransactionsPerSecond(Average) Dimension Data Amazon Web Services IBM SmartCloud Rackspace Linux Cloud Server Local File Performance PostMark 1.51 Benchmark (as reported by Phoronix Test Suite 3.6.1) 2 vCPUs/4 GB RAM1 vCPU/2 GB RAM 4 vCPUs/8 GB RAM Note: Neither IBM nor Rackspace offers a 1 vCPU solution. Default file systems used: ext4 for Dimension Data, ext3 for the other solutions. IBM running RHEL 6.3, all others running Ubuntu 10.04 LTS Server. All systems were 64-bit. Medium LargeSmall
  4. 4. For medium systems, Dimension Data’s throughput 3.26Gbps is 9X times that of Rackspace. For large systems, Dimension Data’s throughput of 4.46Gbps is 9.3X that ofRackspace. Dimension Data’s network throughput in each category is actually greater than or equal to the combined throughput of the othersolutions. IBM is the closest competitor providing close to a full, bi-directional GbE connection in the medium and large scenariostested. Amazon’s medium and large server instances both provide between 1 and 1.2Gbpsofnetworkthroughput.SeeFigure4. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 4 of 7Tolly.com Source: Tolly, May 2013 Figure 4 0 1000 2000 3000 4000 5000 479377233 1,8641,834 1,244 1,052 344 4,463 3,260 2,252 BidirectionalThroughputPerSecond(Avg.Mbps) Dimension Data Amazon Web Services IBM SmartCloud Rackspace Linux Cloud Server Bidirectional Local Area Network Performance Iperf Benchmark (as reported by Iperf 2.0.4) Note: Neither IBM nor Rackspace offers a 1 vCPU solution. For this test, the Rackspace“small”machine has 2vCPUs and 2GB RAM. All tests run with a “large”system as a partner across low-latency, internal network. Throughput can exceed GbE because of 10GbE back-end trunking. IBM running RHEL 6.3, all others running Ubuntu 10.04 LTS Server. All systems were 64-bit. Medium LargeSmall 2 vCPUs/4 GB RAM1 vCPU/2 GB RAM 4 vCPUs/8 GB RAM Founded in 1983, Dimension Data plc is an ICT services and solutions provider that uses its technology expertise, global service delivery capability, and entrepreneurial spirit to accelerate the business ambitions of its clients. Dimension Data offers a versatile suite of cloud services that are capable of meeting the demands of Service Providers, developers, IT managers, or global CIOs. We build superior solutions that take the complexity out of‘the cloud’with developer and IT-friendly cloud services that are fast, secure, easy-to use scalable enough to run complex, high-performance applications.  For more information, visit us at www.dimensiondata.com/cloud. About Dimension Data Source: Dimension Data
  5. 5. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 5 of 7Tolly.com Web/Application Cloud Servers Under Test Virtual Hardware Configuration (as reported by Phoronix Test Suite 3.6.1) Source: Tolly, May 2013 Table 1 Solution Provider System Category Small Medium Large Dimension Data Amazon IBM Rackspace Processor:IntelXeonE7-4830@ 2.13GHz(1Core),Motherboard:Intel 440BX,Chipset:Intel440BX/ZX/DX, Memory:1x2048MBDRAM,Disk:11GB Virtualdisk,Graphics:VMwareSVGAII, Network:Intel82545EMGigabit Processor:IntelXeonE7-4830@ 2.13GHz(2Cores),Motherboard: Intel440BX,Chipset:Intel440BX/ ZX/DX,Memory:1x4096MB DRAM,Disk:11GBVirtualdisk, Graphics:VMwareSVGAII, Network:Intel82545EMGigabit Processor:IntelXeonE5-46500@ 2.70GHz(4Cores),Motherboard:Intel 440BX,Chipset:Intel440BX/ZX/DX, Memory:1x8192MBDRAM,Disk:11GB Virtualdisk,Graphics:VMwareSVGAII, Network:Intel82545EMGigabit Processor:IntelXeonE5430@2.67GHz (1Core),Memory:2048MB,Disk:8GB (m1.smallinstance1EC2ComputeUnit) Processor:IntelXeonE5430@ 2.66GHz(1Core),Memory: 4096MB,Disk:8GB (m1.mediuminstance2EC2 ComputeUnits) Processor:2xIntelXeonE5430@ 2.66GHz(2Cores),Memory:8192MB, Disk:8GB (m1.largeinstance4EC2ComputeUnits) N/A (ThesmallestIBMconfiguration matchedthe2vCPU/4GBRAM “medium”systemcategory.) Processor:QEMUVirtual@2.40GHz (2Cores),Motherboard:RedHat KVM,Chipset:RedHatVirtio, Memory:1x4096MBRAM,Disk: 59GB,Graphics:CirrusLogicGD 5446,Network:RedHatVirtio device Processor:QEMUVirtual@2.27GHz(4 Cores),Motherboard:RedHatKVM, Chipset:RedHatVirtio,Memory: 8192MB,Disk:59GB,Graphics:Cirrus LogicGD5446,Network:RedHatVirtio device Processor:AMDOpteron4170HE@ 2.10GHz(2Cores),Memory:2048MB, Disk:79GB (NextGeneration) (Used for network performance testing only.) Processor:AMDOpteron4170HE@ 2.10GHz(2Cores),Memory: 4096MB,Disk:158GB(Next Generation) Processor:AMDOpteron4170HE@ 2.10GHz(4Cores),Memory:8192MB, Disk:315GB(NextGeneration) Note: Hardware as detected by Phoronix Test Suite may not match virtual hardware environment as documented by the solution provider. Except for IBM, all systems ran Ubuntu (Linux) 10.04 LTS Server (64-bit), Kernel: 2.6.32-33-server. As IBM did not offer that image, IBM ran Red Hat Enterprise Server (Linux) 6.3 (64-bit), Kernel 2.6.32-279.19.1.el6.x86_64.
  6. 6. Test Setup & Methodology Cloud Servers Tolly engineers built all cloud servers under test using publicly available cloud server solutions from each vendor with default configurationsfromeachvendor. Testing was conducted on three categories of servers based primarily on the number of virtual CPUs (vCPUs) provided and, secondarily, on the amount of memory (RAM)provided. The system categories were defined as: 1) Small - 1vCPU, 2GB RAM; 2) Medium - 2vCPUs, 2GB RAM; 3) Large - 4vCPUs, 8GB RAM. For Amazon Web Services, the equivalent Elastic Compute Cloud (EC2) (EC2) unit was used.1 Default file systems were used: ext4 for Dimension Data, ext3 forallothers.SeeTable1. All servers were instantiated in a North American data center for the respective solution. Furthermore, and of importance forthenetworkperformancetests,allcloud servers for a given provider were instantiatedinthesamedatacenter. Except for IBM, all systems under test ran Ubuntu Linux 10.04 LTS Server. Because IBM did not offer this image, IBM servers ranRedHatEnterpriseLinux6.3.SeeTable1. System Resource PerformanceTests PhoronixTest Suite The CPU, RAM and File transaction tests were all run using the Phoronix Test Suite (PTS). PTS is an automated, open-source testing framework. For more information, see:http://www.phoronix-test-suite.com/. PTS version 3.6.1 was used in this test. Virtual hardware shown (in Table 1) as determined by the PTS “Show System Hardware” function. Testers chose tests from the PTR “Complex System Test” category. The tests are automated and repeated automatically until there are sufficient repeated runs with alowstandarddeviationinresults. CPU CPUpowerwasdeterminedbyrunningthe PTS C-RAY 1.1 benchmark. According to OpenBenchmarking.org, the program is a simple ray tracer that tests floating-point CPU performance. A more powerful system will complete this test faster. Thus, a lower numeric result (i.e. shorter run time) equatestoabetterscore. RAM System memory performance was determined by running RAMSpeed SMP 3.5.0(PTSv1.4.0). Local File Input/Output Local file transaction performance was determined by running PostMark 1.51 (PTS v1.1.0).AccordingtoOpenBenchmarking.org, this test is designed to simulate small file tasks typical of web and email servers. File sizesrangefrom5KBto512KB. NetworkThroughput Local area network (LAN) throughput tests were conducted using Iperf, an open source network benchmarking tool availableatSourceForge.net. All network throughput tests were run on pairsofmachineswithonemachinealways running the “large” instance. All machines were configured to be in the same data center and testers confirmed that minimal latencyexistedbetweenthepairsofservers undertest. Tests were run on the private/service networks for AWS, Dimension Data and Rackspace. IBM offers only a single, public addressfortheserverstested. Iperf Iperf is implemented as a command-line, client-server architecture with all data beingtransmittedclient-to-server.Tocreate bidirectional traffic,, each server ran both a client instance and a server instance. Tests were run 5 times with each test set for 60 seconds.2 The server instance was configured: Iperf -s -p 10000 -w 32M&. The client instance was configured: Iperf -c x.x.x.x -p 10000 -t 60 -P 20-w6M,wherethex.x.x.xwasreplacedby theIPaddressofthetargetserversystem. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 6 of 7Tolly.com 1 For a more detailed explanation of EC2, see: http://aws.amazon.com/ec2/instance types/. 2 Because each “end” of the Iperf session calculates throughput relative to its perceived run time, results could vary by approximately 3 5% when viewed on opposite ends of the session connection.
  7. 7. Dimension Data Cloud Server Performance #213131 © 2013 Tolly Enterprises, LLC Page 7 of 7Tolly.com About Tolly The Tolly Group companies have been delivering world-class IT services for more than 20 years. Tolly is a leading global provider of third-party validation services for vendors of IT products, components andservices. You can reach the company by E-mail at sales@tolly.com,orbytelephoneat +1561.391.5610. VisitTollyontheInternetat: http://www.tolly.com 213131-EF1-wt-2013-06-13A-VerI Terms of Usage This document is provided, free-of-charge, to help you understand whether a given product, technology orservice merits additional investigation for your particular needs. Any decision to purchase a product must be based on your own assessment of suitability basedonyourneeds. ThedocumentshouldneverbeusedasasubstituteforadvicefromaqualifiedITorbusinessprofessional. This evaluation was focused on illustrating specific features and/or performance of the product(s) and was conducted under controlled, laboratory conditions. Certain tests may have been tailored to reflect performance under ideal conditions; performance may vary under real-world conditions. Users should run tests based on their own real-world scenarios to validate performance for their own networks. Reasonable efforts were made to ensure the accuracy of the data contained herein but errors and/or oversights can occur.The test/ audit documented herein may also rely on various test tools the accuracy of which is beyond our control. Furthermore, the document relies on certain representations by the sponsor that are beyond our control to verify. Among these is that the software/ hardware tested is production or production track and is, or will be, available in equivalent or better form to commercial customers. Accordingly, this document is provided "as is," and Tolly Enterprises, LLC (Tolly) gives no warranty, representation or undertaking, whetherexpressorimplied,andacceptsnolegalresponsibility,whetherdirectorindirect,fortheaccuracy,completeness,usefulness orsuitabilityofanyinformationcontainedherein.Byreviewingthisdocument,youagreethatyouruseofanyinformationcontained herein is at your own risk, and you accept all risks and responsibility for losses, damages, costs and other consequences resulting directly or indirectly from any information or material available on it. Tolly is not responsible for, and you agree to hold Tolly and its related affiliates harmless from any loss, harm, injury or damage resulting from or arising out of your use of or reliance on any of the informationprovidedherein. Tollymakesnoclaimastowhetheranyproductorcompanydescribedhereinissuitableforinvestment. Youshouldobtainyourown independent professional advice, whether legal, accounting or otherwise, before proceeding with any investment or project related to any information, products or companies described herein. When foreign translations exist, the English document is considered authoritative. To assure accuracy, only use documents downloaded directly from Tolly.com. No part of any document may be reproduced, in whole or in part, without the specific written permission ofTolly. All trademarks used in the document are owned by their respective owners. You agree not to use any trademark in or as the whole or part of your own trademarks in connection with any activities, products or services which are not ours, or in a manner which may be confusing, misleading or deceptive or in a mannerthatdisparagesusorourinformation,projectsordevelopments. Dimension Data Cloud Services For more information about Dimension Data’s Cloud Services, E-mail us at cloud-info@dimensiondata.com or visit us at http://www.dimensiondata.com/cloud.

×