Significance of Standard deviation in Analyzing results

[santosh yeleri]

For two successive load test(after change in the code) for the same application, under same conditions(and all conditions remaining the same) how do we check that the difference(mathematically) in results are because of the changes implemented in the code?

 

I read some material on internet, with respect to this and I found that it can be done with the help of standard deviation, but I was unable to grasp or understand it completly. Please share your understanding with respect to this.

 

Thanks and Regards

Santosh Yeleri

[James Pulley]

Standard deviation is a measure of the spread of your data from the
average point in the data set. Data sets with very small deviations
tend to be clustered around a specific point and are also examples of
processes which scale well, for when resources become restricted under
high loads the performance of the given measured sample varies only
slightly.

Visually, for the frequency distribution of the samples think of the
back half of a normal curve with a very short period with a point at
which the samples are no faster. This "no faster" point occurring
under any load would generally take an architectural reconsideration
of the app to get any faster. This tall skinny, distended curve
with only a few samples outside of this curve window would be an
example of a very optimized business process with a small standard
deviation. Contrast this to many processes which are highly resource
dependent and have wild swings under load where the curve will be long
and flowing, with many hills and valleys in the frequency distribution
of the samples. Your job in tuning the application is to where
possible take this wild undulating curve and through platform tuning
and code changes push the frequency curve to the left. coalescing as
many data points as close to the average value as possible.

That's the layman's 10000 foot view of what standard deviation means
to you and how to visualize it in your data.

it is very worthwhile to pick up your dusty copy of your college stats
text and go back through the material. If you have not taken a
statistics course then you might want to consider one of the "Dummies"
series. While I have not gone through "Statistics for Dummies" I
have a whole bookshelf of bright yellow books in my office that have
been great references over the years, so based upon that alone I would
have no problem recommending another friendly yellow book.

James Pulley, http://www.loadrunnerbythehour.com/PricingMatrix

[santosh yeleri]

Thanks for this information James

--
You received this message because you are subscribed to the Google "LoadRunner" group.
To post to this group, send email to LR-Loa...@googlegroups.com
To unsubscribe from this group, send email to
LR-LoadRunne...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/LR-LoadRunner?hl=en