8834 Capital of Texas Highway North Austin, Texas 78759 Voice 512.652.3500 Fax 512.652.3501 Toll Free 866-652-3500
P e r f o r m a n c e W h i t e P a p e r
6
3. Performance Methodology Details
To make this paper applicable to many types of busi-
nesses, ClearCube simulated and measured perfor-
mance in different congurations. This section com-
municates those results.
a. Simulating an Ofce Worker in a Corporate
Environment
A standard baseline for most corporate congura-
tions is an ofce worker who interacts with differ-
ent types of documents throughout the workday.
While some virtualization studies have focused on
the use of Microsoft Word as the sole application for
ofce use, this may be too simple an approach. In
this study, we simulated a standard workload that
uses multiple Microsoft Ofce applications. This load
more realistically simulates a typical ofce worker’s
interaction and load on a system. Performance loads
used include:
• Microsoft Word
• Microsoft PowerPoint
• Microsoft Excel
To make the performance measurements effective
and repeatable, a scripting tool simulated a user in-
teracting with the computer. A script, written to auto-
matically create and edit documents, accomplished
the following:
• Create a document
• Add content to the document at the speed
of a normal person typing and using a
mouse to interact with the system
• Edit the document to make changes
• Save and close the document
• Repeat• Repeat
The script created multiple documents of each type
(Word, PowerPoint, and Excel). Each document was
created and left open while other documents were
subsequently created and edited. This simulated a
user interacting with different documents, and leav-
ing some of them open for use later in their work
session. After the multiple document editing was
complete, the script closed all documents.
The script was setup to run multiple times until can-
celled. In this way, the script was running on multiple
VMs and on each VM simultaneously.
b. Performance Measurement and System
Conguration
To simulate a full set of VMs on a fully loaded sys-
tem, it is insufcient to simply install the software
and run the performance test only on the VMware
console. It is necessary to install a fully congured
system as shown in Fig 2. ClearCube used I/Ports
for the user component to ensure the IP network
trafc emulated people typing and interacting with
the entire system.
Consider the conguration of the A1010 blade with 3
GB RAM. An I/Port was congured and established
connection with each VM during the conguration
testing. To determine how many VMs were possible
on this system, the performance was measured suc-
cessively with the following VMs and I/Ports added:
• One VM using one I/Port; no other I/PortsOne VM using one I/Port; no other I/Ports
connected to the system
• Two VMs using two I/Ports (one VM perTwo VMs using two I/Ports (one VM per
I/Port); no other I/Ports connected
• Three VMs using three I/Ports (one VMThree VMs using three I/Ports (one VM
per I/Port); no other I/Ports connected
• Four VMs using four I/Ports (one VM perFour VMs using four I/Ports (one VM per
I/Port); no other I/Ports connected
• Five VMs using ve I/Ports (one VM perFive VMs using ve I/Ports (one VM per
I/Port); no other I/Ports connected
• Six VMs using six I/Ports (one VM perSix VMs using six I/Ports (one VM per
I/Port)
This testing methodology is represented in the fol-
lowing gure.
Performance measurements were taken for each
Manymanuals.com
Manymanuals.de
Manymanuals.fr
Manymanuals.it
Manymanuals.pl
Manymanuals.cz
Manymanuals.es
Manymanuals-pt.com
Kommentare zu diesen Handbüchern