CODEWITHSUNDEEP
loadrunner
PrashantMP
PrashantMP

Reputation: 11

Why is auto correlation in loadrunner not always accurate?

I am still not able to understand why the auto correlation in loadrunner is not always accurate? Why it is unable to identify all the correlation qualifiers? For example, it will try to correlate small variables like time but will miss out the big sessiodID variables?

Upvotes: 0

Views: 1754

Answers (2)

Speedmaster
Speedmaster

Reputation: 21

"I am still not able to understand why the auto correlation in loadrunner is not always accurate? Why it is unable to identify all the correlation qualifiers?"

I like to use the analogy of a grammar check in a word processor. A lot of what it does is "soft" and somewhat subjective.

Upvotes: 0

James Pulley
James Pulley

Reputation: 5692

Well, the autocorrelation rules are very generic. As a result small changes in the application source will often render the rules obsolete in terms of ordinal or LB/RB conditions. You also have some variables which are often larger than the pattern match buffer for the autocorrelation egine, such as state management variables in asp and asp.net environments. In such a case the LB is easy to identify but the RB condition is never located because it is off the end of the pattern match buffer.

Autocorrelation is great in theory and there are some environments where it works very well, such as for the stararray structure in siebel environments, but most of the other time you should consider these items starting points to "tune" the recommendation so you receive a constant match every time with a strong left and right boundary condition instead of looking the the fifteen ordinal of the LB condition of the equals symbol with quote for right boundary. If you like horror films then the following analogy will make you smile. Autocorrelation is like a silver bullet. In the off chance that you run up against a werewolf it is invaluable. Otherwise it is just a very expensive bullet and you still need to be able to use the gun to aim and fire effectively.

Many organizations treat autocorrelation as a direct substitute for training people on the subject of manual correlation. One is not a direct substitute for the other. It is nearly an axiom that if you are not proficient with manual correlation then you will be left blaming the tool for being ineffective when autocorrelation fails....when in reality this is a skills gap.

Lastly, here is a podcast link that should help

http://www.perfbytes.com/dynamic-data-correlation

Upvotes: 2

Related Questions