Java Performance Services
Training, Seminars, Benchmarking, Tuning

Java Performance Tuning Course


Chania Crete, May 17-20, 2010


Sun Extreme Learning EXL-2025

Houston, December 1-4,2009
New York, December 8-11, 2009
Washington DC, January 5-8, 2010



San Francisco, January 11-14

Anti-if

I have joined Anti-IF Campaign

Calendar

««Nov 2009»»
SMTWTFS
1234
5
67
891011121314
15161718192021
22232425262728
2930

Performance Anti-Patterns

My Top Tags

                                       

Mailing List

My RSS Feeds








The User Experience is is King

« H E email
posted Sunday, 24 August 2008

I was just reading an article entitled Testing for Performance. In the sidebar the editor pulled out this quote. "When modeling, remember that not all performance testing is focused on the end-user response times. Sometimes there are better measures than end-user response time."

I'll have to say that I (almost violently) object to this statement. There is nothing more important than the end-users time and if you're not focused on that than you are doing your clients a disservice. In my books, end-user experience is king and if there is anything that will destroy the user experience it is poor response times. IMHO, your goal is to waste as little of the users time as is possible given your hardware/budget constraints.


comments (3)


1. William Louth left...
Monday, 25 August 2008 8:32 am :: http://williamlouth.wordpress.com

Hi Kirk, I strongly object to your "violent" objection. You seem to miss the point that there our trade offs to be made and focusing solely on the end user response time can result in the team losing site of the scalability requirements of the application. XPE and SPE recognize this and also recognize that you cannot please all the users all the time. You need to balance the needs of many different parties and user groups. It would be silly and naive to attempt to reduce every single use case response time as each tune-up could have the potential to invalidate previous gains and you would be in a endless loop. Have you every considered (or seen) cases were the speedy execution of one particular use case has a negative knock on effect to others executing concurrently or shortly thereafter. The focus is still on the end user response times but in an responsible and intelligent manner.


2. Kirk left...
Monday, 25 August 2008 12:00 pm

Hi William,

I would have been disappointed if you'd not have objected.

I would think that looking after scalability issues would help end user response times. If they don't I'd have to question what you're doing. Also, I never suggested that you should tune every single interaction so make it run as fast as possible. That would violate my budget/resource caveat. So yes, you are right when you say end user response times in a responsible and intelligent manner. However, the pulled quote didn't quite suggest that.


3. Markus Kohler left...
Monday, 25 August 2008 8:40 pm :: http://kohlerm.blogspot.com/

Hi all, I agree in general that response time is king. I remember that I've seen a slide by Google where they said that a a few hundred milliseconds cost them 20% (or so) page clicks.

When looking at the details, there are operations where users don't expect a response time of less then a second. Therefore it might not hurt that some "pages" load slower.

In practice "perceived" performance is really the king. The sooner the user has enough information on the screen to be able to proceed, the better. For web pages that doesn't mean for example that all images have to be loaded before the user can proceed.

Regards, Markus