snehadri
Junior Member
Reged: 07/19/06
Posts: 4
Loc: Kolkata
|
|
Can any one of you tell me what is the exact working formula for 90 percent response time in loadrunner analysis part?
thanx in advance..............
|
Skiing Lizard
Member
Reged: 03/22/06
Posts: 71
Loc: Belgium
|
|
yeah it's pretty easy...
line all the response times for a transaction up in a row ordered from low to high... then go to the 90th in the line...
that equals the 90th percentile
if there are 50 then it's the 45th if there are 200 then it's the 180th etc
|
cmeisenzahl
Super Member
Reged: 07/22/02
Posts: 1437
|
|
As an aside, I prefer to use 95% instead of 90%.
Chris
--------------------
Interested in economics, fountain pens, watches? Check out:
http://www.pretenseofknowledge.com/
http://twitter.com/speedmaster
|
m_adcock
Member
Reged: 06/07/01
Posts: 97
Loc: New York, NY USA
|
|
Another way to think of this metric when explaining it to someone is to say that the application's end-users could expect to encounter this response time (or less) 90% of the time.
--------------------
Regards,
Matthew Adcock
RTTS - The Software Quality Experts
Performance Division
Blog: http://www.rttsweb.com/interactive/blogs/blogs.jsp?bc=Matthew%20Adcock&bun=madcock
|
nsc
Member
Reged: 11/13/03
Posts: 30
Loc: USA
|
|
I like to correct Matthew's response...
Another way to think of this metric.. When explaining it to someone is to say that the 90% of the application's end users could expect to encounter this response time (or less).
|
DanCorey
Member
Reged: 08/09/02
Posts: 262
|
|
Restating the comment but not correcting it does not quite make a correction in my book.
If you don't mind, can you please explain the flaw in Matthew's statement?
If I understand it, the 90th percentile (put simply) basically just gets rid of the statistical outliers? And it only gets rid of the longest responses, not the shortest, right?
At least that's my understanding...
--------------------
-- Dan
Loadrunner information link
|
JakeBrake
Moderator
Reged: 12/19/00
Posts: 15290
Loc: St. Louis - Year 2025
|
|
90% of that transactions's response times were at or below the stated value.
This metric is useful in determining acceptability, since one cannot guarantee 100% of anything in the world of the web.
|
DanCorey
Member
Reged: 08/09/02
Posts: 262
|
|
Ok, cool. Pretty much what I always thought it meant.
Thanks for the confirmation Jake.
--------------------
-- Dan
Loadrunner information link
|
miya
Junior Member
Reged: 07/27/04
Posts: 9
Loc: L'viv
|
|
Guys
It is clear that 90% line is 90% of requests. But how 10% that are not included to results are counting?
What i mean:
For example, we have 10 requests with such response time:
1st - 10ms, from 2nd till 8th let it be - 12ms, 9th - 20ms, 10th - 30ms
so what is gonna be 90% line?
1 case: (10+12*7)/8
2 case: (12*7+20)/8
For my opinion correct is 1 case, but which formula is used in performance tools?
|
Ferdie
Junior Member
Reged: 09/06/04
Posts: 176
Loc: Tamil Nadu, India
|
|
The 90th percentile is a measure of stastical distribution, not unlike the median. The median is the middle value. The median is the value for which 50% of the values were bigger, and 50% smaller. The 90th percentile tells you the value for which 90% of the data points are smaller and 10% are bigger.
Statistically, to calculate the 90th percentile value:
1. Sort the transaction instances by their value.
2. Remove the top 10% instances.
3. The highest value left is the 90th percentile.
Example:
There are ten instances of transaction "t1" with the values 1,3,2,4,5,20,7,8,9,6 (in sec).
1. Sort by value -- 1,2,3,4,5,6,7,8,9,20.
2. Remove top 10 % -- remove the value "20."
3. The highest value left is the 90th percentile -- 9 is the 90th percentile value.
The 90th percentile value answers the question, "What percentage of my transactions have a response time less than or equal to the 90th percentile value?" Given the above information, here is how LoadRunner calculates the 90th percentile.
--------------------
Performance Analyst
http://www.csscorp.com
|
Previous
Index
Next
Threaded
Edit
Reply
Quote
