Testing

Hardware Configuration:

  Web Front End 1 Web Front End 2 Application Server Database Server

Server Model

Dell PowerEdge 1850

Dell PowerEdge 1850

Dell PowerEdge 1850

Dell PowerEdge 6800

Processors

2 Intel Xeon Irwindale 2.8GHz

2 Intel Xeon Irwindale 3.0GHz

2 Intel Xeon Irwindale 3.0GHz

 (4) Intel Xeon 7140M

3.4GHz

RAM

8GB

8GB

10GB

64GB

HardDrive

73GB 10K SCSI

73GB 10K SCSI

73GB 10K SCSI

73GB 15K SCSI

Software Configuration:

  Web Front End 1 Web Front End 2 Application Server Database Server

Operating System

Windows Server 2008 R2 SP1

Windows Server 2008 R2 SP1

Windows Server 2008 R2 SP1

Windows Server 2008 R2 SP1

Software

SharePoint 2010, Social Sites 2010 v1.2 (aka Skyvera Social)

SharePoint 2010, Social Sites 2010 v1.2 (aka Skyvera Social)

SharePoint 2010, Social Sites 2010 v1.2 (aka Skyvera Social)

SQL Server 2008 R2

UAC

Disabled

Disabled

Disabled

Firewall

Disabled

Disabled

Disabled

Methodology:

A series of test were performed for both the SP and AS page to determine the maximum VSTS load while maintaining the Green Zone constraints.

We performed tests for the following SharePoint configurations: 1x0x1 (1 WFE and AS, 1 DB), 1x1x1, and 2x1x1. The 1x1x1 and 2x1x1 environments were configured in a SharePoint farm environment with the web components enabled for the WFEs and the application services disabled, while the WFEs have all application services enabled and the web components disabled.

Before performing the simulated "production" load a number of tests were performed to add content to the activity stream. After determining the maximum VSTS load an additional test was performed to gather the results used in this guide.

Notes:

The tests were performed without load balancing hardware or software. During test we found that load balancing had the potential to skew the results by shifting more traffic to particular WFEs (stickyness).

Also there is slight hardware variations between the machines used at WFEs. The performance test where tailored to provide equal traffic to all web frontends and a total was taken for the combined throughput of all WFEs.