On Mar 13, 2014, at 11:08 AM, Stéphane Ducasse <
[hidden email]> wrote:
“...Because of the doubts about scalability and the persistence issues, it is that when doing web apps I use pharo to develop and GemStone to deploy :-).”
persistance is a bottleneck waaaaay before Pharo so I’d do load tests to have something measurable to hook some confidence into.
What comes to mind is
1. to have a set of load tests from Pharo to the database
2. to have a set of load tests from JMeter to the Pharo web app (actions without involving database*)
3. another set of load tests for the full stack
It’s tedious but you need to build self confidence on your specific stack
Best part?
Is not faith, is the numbers you get (so is testable and demonstrable)
Here is some JMeter tutorials:
When I did some, I remember getting something in the range of 40 to 80 requests per second per worker image (no db involved)
have fun! don’t forget to share what’s possible!
*that might or might not be possible depending on the web app. For me, in airflowing, every http request is a transaction (using a seaside filter for that)
_______________________________________________
Pharo-business mailing list
[hidden email]
http://lists.pharo.org/mailman/listinfo/pharo-business_lists.pharo.org
Locked
