|
Re: Hardware recommendations for RAP server [message #1060154 is a reply to message #1059199] |
Thu, 23 May 2013 10:59 |
|
Hi Ken,
the CPU load of a RAP Server depends heavily on the application so there
are hardly any general recommendations, except: test it! The factors
that influence the performance most are:
* number of widgets: CPU load depends on the total number of widgets in
your UI, not only visible ones (this might change in one of the next RAP
releases)
* request frequency: lots of modify listeners, drag&drop, etc. lead to
much more requests than rather static UIs
* layouts: mote dynamic layouts require much more calculation than more
static ones (e.g. text size determination)
* the operation mode: JEE performs better than SWT compatibility mode
In our recent load tests, 1000 sessions performed well on a single
server under heavy load. Depending on your application, a powerful
server may or may not handle 2000 parallel user sessions.
You should definitely perform your own performance tests with your
actual application to answer this question. Using cloud hardware you can
easily test with different server configurations.
Regards,
Ralf
--
Ralf Sternberg
Twitter: @EclipseRAP
Blog: http://eclipsesource.com/blogs/
Professional services for RAP and RCP?
http://eclipsesource.com/services/rap/
|
|
|
|
Powered by
FUDForum. Page generated in 0.07252 seconds