Hi Gabriel,
In brief answer to your questions, I'm interested in what might be
causing the slowness of your app. Have you tried looking at the (Apache)
error log to see the SQL queries it's generating? Perhaps Links has
failed to appropriately compile your queries--this is a common
situation. Sometimes we can rewrite the code to make it produce a good
query; other times it's simply a bug that we can fix in Links.
It's also possible that the big source file is causing some of the
trouble, but I'd be surprised if it was making that much difference. If
it turns out not to be the database queries, then we can look at other
causes.
Cheers,
Ezra
G Tellez wrote:
Dear links users,
Just for curiosity, I've been conducting some tests (outside the
University network) using NeoLoad (http://www.neotys.com/), a tool for
load and stress testing based on test recording. The tool simulates
virtual users collecting web server data based on request & response.
Tests can be parameterised, for example, change user's bandwidth or
simulate "ramp up" scenarios, etc.
I've tested 2 applications with an initial quantity of 2 users, adding
2 new users every 2 minutes in a total time period of 10 minutes (10
virtual users in total, the maximum number with an evaluation license)
with the following results:
- linksCollab (Steve's application, located at
http://linkscollab.stevestrugnell.net/). Simulating an usual
navigation session: displaying initial page, logging-in, selecting a
project, viewing a milestone, consulting the user's profile.
--> Average page response time is 8.64s.
- LinksPetStore (an e-commerce application, located at frege server).
Simulating an usual navigation session: displaying initial page,
consulting a pet category, consulting a product, displaying product's
details, creating a shopping cart, adding the item to the shopping
cart, checking-out, logging-in, confirm shipping/billing addresses,
and receiving confirmation.
--> Average page response time is 25.1s
linksCollab is a web application splitted in several files, using the
include feature. LinksPetStore is an e-commerce application, (coded in
a single file; 2,600 LOC approximately), whose shopping cart status is
kept in the DBMS. So each time a user adds a new item/updates the
cart, the application has to go and retrieve the info from the database.
My doubts are:
- Taking into consideration that Links has to be compiled into
JavaScript, the average page response time could be that high in the
LinksPetStore because it's coded in a single source file?
- Or it could be because of the continuous access to the database?
- Has anyone used a testing tool like this one before?
- Based on the description of the tests, can you suggest a special
test case to try?
Thank you very much,
Gabriel
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
_______________________________________________
links-users mailing list
[email protected]
http://lists.inf.ed.ac.uk/mailman/listinfo/links-users
