loadrunner which recording mode(HTML or URL) is faster...

[SP Manjunath]

which recording mode(HTML or URL) is faster execution?share me exact examples for URL recording needs?can u pls ans

 

[James Pulley]

Considering that your pacing model per user would be identical across recording models, none of them should be faster than the other (because of the think time and pacing). The resource fingerprint per virtual user does change by recording type. All as web_custom_request() is going to have the smallest resource fingerprint on the load generators per virtual user, Truclient the most. This translates to how many users you can achieve per host before your load generator slows and colors your test results

 

[Deen John]

URL mode is faster as per my understanding..

 

[Deen John]

in url.mode each req is recorded explictly..making faster for replay

 

[James Pulley]

Your hypothesis is faulty. All the same speed with think time and pacing in place. Different resource costs.

 

[Vidya Khandagale]

I think script will run faster if application is fast with any of the recording type html or url. There could be chage in load generator resources like cpu and memory with the type of protocol you use and type of application. URL mode records all hidden data like session id, css, jpeg, jpg files and record separate requests for all the images, css regardless to user clicks context. Most of the times if script does not work with html mode we can always check if it works with html advance explicit url option where we get web_submit_data instead of web_submit_form. It works most of the cases with accurate correlations. If it still fails go for URL mode.

 

[Gaurav Nayyar]

URL mode will be faster !!!! As HTML mode records in Context Sensitive Mode which includes dynamic parsing which leads to higher Memory consumption and resource utilization and at the same time effects script execution Time as well.

 

[James Pulley]

Record one page and time it in all modes. Be sure to include pacing between requests. Run these in the controller (not vugen) for one user on a load generator. Collect 1000 samples of each. Compare

 

[Petar Puskarich]

Both modes have their pros and cons - know your application very well and the choice becomes clear. I would NOT want to do extremely dynamic sites in URL mode. Well, unless you really like to rewrite the parser used in html mode for fun Once you start adding in the extra pieces to take care of things non wire driven then you end up back at html mode 90% of the time. The overhead cpu wise is not that much really. On properly calibrated and tuned Win 2k3 32 bit machines we were pushing about 1200 vusers per old school Dual PIII 800mhz Compaq DL360 G1 servers. I hand tuned and optimized those os builds for maximum efficiency on the io, memory, and cpu execution cycles needed however.

 

[Chidvilash Vakada]

URL mode will be faster than HTML mode records but in complex drive mode recording it gives same results at the time of script execution

 

[James Pulley]

What good would a performance testing tool be if different recording modes produced different response times, I.e. one model is faster than another? I can just hear those challenges from developers where you have just called their (code) baby ugly, "I reject your results that imply that my code is slow, when everyone knows that if you had just recorded in url mode it would be faster and everything would be in spec. You need to rebuild your test in URL mode and re execute it before your results will be accepted. Well, next time I will reject it because you have told us that one user represents n real users where you have eliminated think time and pacing"

 

[Gaurav Shukla]

Resource utilization is certainly affected by choice of vuser type. As far as recording mode is concerned LR certainly does more work in HTTP/HTML mode, but if you are talking abt replaying it...well thts precisely controlled by pacing/thinktime, and if we are here talking abt gut feeling...well then choose wteva u feel like