Vconsolidate Performance and Power on IBM and HP Quad-Core

TEST REPORT SUMMARY vConsolidate performance and power on IBM and AUGUST 2008 HP quad-core multiprocessor Intel processor-based servers

IBM Corporation (IBM) commissioned Principled Technologies (PT) to KEY FINDINGS measure performance and power with Intel’s vConsolidate OEM z Higher performance (see Figure 1) version 1 workload (profile 2) using VMware ESX Server 3.5 update 1 o The 4P IBM System x3850 M2 server delivered on the following quad-core servers: 11.8 percent more performance than the 4P HP ProLiant DL580 G5 server • HP ProLiant DL580 G5 with four 2.93GHz Intel Xeon X7350 o The 8P IBM System x3950 M2 server delivered processors and 32 2GB DIMMs 118.4 percent more performance than the 4P HP • IBM System x3850 M2 with four 2.93GHz Intel Xeon X7350 ProLiant DL580 G5 server processors and 32 2GB DIMMs z Higher performance per watt (see Figure 2) • o The 4P IBM System x3850 M2 server produced IBM System x3950 M2 with eight 2.93GHz Intel Xeon X7350 29.3 percent better performance per watt than processors and 64 2GB DIMMs the 4P HP ProLiant DL580 G5 server o The 8P IBM System x3950 M2 server produced All three systems had identical PCI-e NICs and HBAs, but the IBM 35.8 percent better performance per watt than System x3950 M2 had twice as many of them as the other servers. the 4P HP ProLiant DL580 G5 server Figure 1 provides a normalized comparison for the test servers with z Lower power consumption (see Figure 2) the optimum vConsolidate work units, which it calls consolidation stack o The 4P IBM System x3850 M2 server consumed units (CSUs). This chart normalizes the results to the performance of 13.6 percent less power than the 4P HP ProLiant DL580 G5 server the HP ProLiant DL580 G5. That system’s score is thus 1.00. o The 8P IBM System x3950 M2 server consumed Normalizing makes each data point in the chart a comparative 19.6 percent less power than two of the 4P HP number, with higher numbers indicating better performance. ProLiant DL580 G5 servers would have used

In this summary, we discuss the best results for all servers. For complete details of the performance of each server at peak CSUs, see the test report at www.principledtechnologies.com/Clients/Reports/IBM/IBMvCon0808.pdf.

5 CSUs vCon relative performance Figure 1 shows the IBM x3950 M2 server delivered results at peak CSUs better overall performance than the other two servers. 2.50 It offers 118.4 percent better performance than the HP ProLiant DL580 G5 server and 95.4 percent better performance than the IBM x3850 M2 server. 2.00 10 CSUs We tested the servers with redundant power supplies HP ProLiant DL580 G5 active. As Figure 2 illustrates, the IBM x3950 M2 1.50 IBM x3850 M2 server delivered the highest performance per watt on IBM x3950 M2 the 10-CSU vConsolidate workload, a 35.8 percent increase over the HP ProLiant DL580 G5 server at 1.00 five CSUs and a 5.0 percent increase over the IBM Normalized score 5 CSUs 5 CSUs x3850 M2 server also at five CSUs. 0.50 We calculated performance per watt by dividing the vConsolidate score for each server at peak CSUs by 0.00 the measured power when running at peak CSUs for Server/number of CSUs a minimum 30-minute interval. We measured power at 208 V on all three servers. Figure 1: vConsolidate results at the optimal number of CSUs for the three servers we tested.

Normalized Normalized performance Server vCon results Average power Idle power performance per watt HP ProLiant DL580 G5 (5 CSUs) 2.72 1.00 890.3 641.5 1.00 IBM x3850 M2 (5 CSUs) 3.04 1.12 769.5 514.0 1.29 IBM x3950 M2 (10 CSUs) 5.94 2.18 1,431.3 1,016.2 1.36

Figure 2: A comparison of the three servers in performance, power, and performance per watt at the peak number of CSUs.

For more information on these tests and to see the full test report, visit: www.principledtechnologies.com/clients/reports/IBM/IBMvCon0808.pdf.

Principled Technologies, Inc. 1007 Slater Road, Suite 250, Durham, NC 27703 www.principledtechnologies.com 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.