by Edison Group, Inc, For Intel April 4, 2006

Introduction

Purpose of the Study

Edison Group undertook the study described in this report to assess the scalability of Intel®-based 2-way and 4-way server platforms in typical applications in consolidated server environments. In the absence of any industry-standard virtualization capacity benchmarks, this study provides the reader with a guide to the performance differences between Intel® Xeon® 2-way and 4-way servers using their entry-level and high-end processor options for virtualized solutions. The results are presented in this report.

Who Should Read This Report

This report should assist IT professionals in deploying or developing a strategy for consolidated and virtualized enterprise servers. The results of this study should help readers decide which virtualized platform best meets their needs by providing data on performance and power scalability between the most popular Intel®-based server types and processors.

Goal of This Study

Improvements in distributed computing technology — particular in operating system design — have made it feasible today to operate multiple server applications as “virtual machines� (VMs) within a single physical platform using Intel® X86 processors. No longer is it necessary to have a server entirely dedicated to e-mail, for example, or a web server, or any shared database or database application.

Server consolidation is not only feasible but is frequently highly desirable. Reducing 100 physical servers to 50, for example, has obvious factors to recommend it in terms of space, maintenance, licensing, etc. Consolidating servers can also reduce the wasteful processing inefficiency of dedicated servers, which commonly operate at well under 20 percent processor utilization.

In light of this and the trend favoring server consolidation, Edison Group set out to measure the effect that higher-performance Intel® 2-way and 4-way servers would have on VM capacity, testing the premise that better platforms and higher-end1 processors will result in greater consolidation capacity. Specifically, tests would be designed to determine the ratio of processing cost to performance strictly in terms of capacity to reduce overall server hardware — for example, if a 20 percent higher-priced processor might yield 60 percent more VM capacity. ROI and TCO factors — while certainly cost considerations for a real-life IT department — are not covered in this study

Therefore, this study strictly helps provide relative VM capacity for the different platforms and/or CPUs in a virtualized environment, without considering other valid platform decision criteria beyond capacity. Edison Group is currently developing a follow-on study to address these other, factors.

Approach

Edison Group wanted to compare the effect of virtual machine density on the highest-volume Intel®-based server platform types (2-way and 4-way), as well as entry-level versus top-end Intel® processors supported on each platform. The approach taken was to replicate the common solutions IT managers are reportedly deploying when they virtualize and consolidate, and then to test the impact on the capacity of targeted Intel®based server platforms with different CPUs. The results provide an indicator of the capacity scaling between Intel®-based server platforms and — within the same platform — of the entry-level versus top-end processors.