16.01.2013 Views

Microsoft Sharepoint Products and Technologies Resource Kit eBook

Microsoft Sharepoint Products and Technologies Resource Kit eBook

Microsoft Sharepoint Products and Technologies Resource Kit eBook

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

230 Part III: Planning <strong>and</strong> Deployment<br />

You can see that the RPS of the portal site stops growing at around 22 RPS. The<br />

other thing that is important to notice is that the latency increases as more load is<br />

generated.<br />

Generating Load <strong>and</strong> Interpretation of the Results<br />

The purpose of stress testing is to determine how a portal site will perform <strong>and</strong> to<br />

identify bottlenecks under load. Running scripts for the different scenarios that accurately<br />

simulate the expected production traffic will generate load on the servers. The<br />

purpose of a performance test should be to find out the maximum throughput under<br />

load. To better underst<strong>and</strong> what might be acting as the bottleneck of the portal site,<br />

it makes sense to check for several counters on the different servers.<br />

The most interesting point to find out is the throughput capabilities of the portal<br />

site. Throughput is the amount of work that the portal site can do in a given time<br />

period. In this case, that would mean finding out how many requests per second the<br />

portal site can h<strong>and</strong>le. Splitting the performance counters into three groups makes<br />

sense. First of all, we have to make sure that the clients are not suffering too much<br />

under the load they are generating. Second, using different counters for the SQL<br />

server <strong>and</strong> the front-end servers will show us where the bottleneck lies (that is, at<br />

the front end or back end).<br />

All the different scripts can be run with the ACT test tool, <strong>and</strong> there will be two<br />

kinds of results that can be analyzed. One type is the Web server responses that tell<br />

you whether there were any HTTP errors, how many pages have been processed,<br />

how long it took to open up the page, <strong>and</strong> so on. The other results are the performance<br />

counters that also indicate server hardware-related results.<br />

Types of Scripts<br />

As defined earlier, the most basic script contains just one line:<br />

Test.SendRequest(“http://sps”)<br />

What this script does is send a request to the server http://sps <strong>and</strong> requests the<br />

default page for the site. Because this is a corporate portal site, the default page is<br />

Default.aspx. The Web server processes the Default.aspx page <strong>and</strong> returns the result<br />

to the client computer. ACT times this whole transaction <strong>and</strong> uses that information<br />

to generate RPS data. Running this script will generate performance data for your<br />

server or server farm, but it is not particularly useful unless the only thing your users<br />

do is go to the Home page of the corporate portal site.<br />

Using a mixed-scenario script is much more useful for testing the portal site.<br />

The following script performs the general workload mix on a specified corporate<br />

portal site. This script can be modified for divisional portal sites or to include other<br />

sites. It can also be modified for specific workload mix requirements of your organization.<br />

The code for the mixed scenario script is as follows:

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!