Re: Is it possible to avoid serialization of page after ajax request?

[Fridolin Jackstadt]

Issue opened here: https://issues.apache.org/jira/browse/WICKET-5933

--
View this message in context: 
http://apache-wicket.1842946.n4.nabble.com/Is-it-possible-to-avoid-serialization-of-page-after-ajax-request-tp4671295p4671331.html
Sent from the Users forum mailing list archive at Nabble.com.

-
To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
For additional commands, e-mail: users-h...@wicket.apache.org

Re: Is it possible to avoid serialization of page after ajax request?

[Martin Grigorov]

Please create a ticket and attach the quickstart there.
Thanks!

Martin Grigorov
Freelancer. Available for hire!
Wicket Training and Consulting
https://twitter.com/mtgrigorov

On Tue, Jun 23, 2015 at 1:31 PM, Fridolin Jackstadt 
wrote:

> Here is a quickstart
>  >
> that demonstrates the performance impact. There is a
> ExperimentalPageManagerProvider configured in the WicketApplication that
> makes ajax polling requests about two times faster.
>
> Something similiar has been done in
> https://issues.apache.org/jira/browse/WICKET-3667, but rolled back. Looks
> to
> me like it's worth to improve.
>
> --
> View this message in context:
> http://apache-wicket.1842946.n4.nabble.com/Is-it-possible-to-avoid-serialization-of-page-after-ajax-request-tp4671295p4671326.html
> Sent from the Users forum mailing list archive at Nabble.com.
>
> -
> To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
> For additional commands, e-mail: users-h...@wicket.apache.org
>
>

Re: Is it possible to avoid serialization of page after ajax request?

[Fridolin Jackstadt]

Here is a quickstart
  
that demonstrates the performance impact. There is a
ExperimentalPageManagerProvider configured in the WicketApplication that
makes ajax polling requests about two times faster.

Something similiar has been done in
https://issues.apache.org/jira/browse/WICKET-3667, but rolled back. Looks to
me like it's worth to improve.

--
View this message in context: 
http://apache-wicket.1842946.n4.nabble.com/Is-it-possible-to-avoid-serialization-of-page-after-ajax-request-tp4671295p4671326.html
Sent from the Users forum mailing list archive at Nabble.com.

-
To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
For additional commands, e-mail: users-h...@wicket.apache.org

Re: Is it possible to avoid serialization of page after ajax request?

[Fridolin Jackstadt]

the page is indeed not deserialized, but received from http session, but it
looks like we cannot avoid the serialization.

Our application doesn't touch the page, but the AbstractPageManager does:
*
IManageablePage page = getRequestAdapter().getPage(id);
if (page != null)
{
getRequestAdapter().touch(page);
}
*

Therefore the requested page is always touched, but we couldn't figure out
why. If it's possbile not to touch the page here (in some cases) that would
solve the problem.

--
View this message in context: 
http://apache-wicket.1842946.n4.nabble.com/Is-it-possible-to-avoid-serialization-of-page-after-ajax-request-tp4671295p4671318.html
Sent from the Users forum mailing list archive at Nabble.com.

-
To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
For additional commands, e-mail: users-h...@wicket.apache.org

Re: Is it possible to avoid serialization of page after ajax request?

[Martin Grigorov]

Hi,

Check
http://ci.apache.org/projects/wicket/guide/6.x/guide/internals.html#pagestoring

The latest used page per user session is kept in the http session. It is a
*live*, detached but not serialized java object.
If Wicket cannot find it there then it will look for a serialized version
of the page in the disk store.

Later a page is stored to the page storages only if it was touched (see
org.apache.wicket.page.IPageManager#touchPage) during the request cycle.

So I guess in your case you have many writes to the disk because the page
is loaded from the http session and then touched due to some logic in your
application.

Please check the user guide, see what exactly happens in your app and give
more details so we can find a solution.

Martin Grigorov
Freelancer. Available for hire!
Wicket Training and Consulting
https://twitter.com/mtgrigorov

On Mon, Jun 22, 2015 at 3:11 PM, Fridolin Jackstadt 
wrote:

> We run into a disc and cpu bottleneck on a wicket page with 1+ users,
> that uses ajax updates via polling (AjaxSelfUpdatingTimerBehavior).
>
> The result of our investigation is that page store needs about 10ms for
> serialization, while the framework serializes and deserializes the same
> data
> again and again.
>
> The question is: is it possible to use a kind of "copy on write"-strategy
> in
> IPageStore, that keeps always the latest verison of the page in memory?
>
> At least for non-versioned pages it should be possible, but IManageablePage
> does not expose if the page is versioned or not. For versioned pages it's
> maybe not possible, because modifications are applied before the PageStore
> has a chance to make a copy of the page.
>
> --
> View this message in context:
> http://apache-wicket.1842946.n4.nabble.com/Is-it-possible-to-avoid-serialization-of-page-after-ajax-request-tp4671295.html
> Sent from the Users forum mailing list archive at Nabble.com.
>
> -
> To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
> For additional commands, e-mail: users-h...@wicket.apache.org
>
>