Google App Engine

Has anybody got Rapid Miner working with Google App Engine? Would it be possible and what would be main considerations? Many thanks.

Find more posts tagged with

AI Studio

Accepted answers

All comments

land

Hi,
as far as I know, Java Programs can't create new threads inside the APP Engine. It will be hard managing more complex tasks without RapidMiner Processes, which are based upon thread creation...

Greetings,
Sebastian

aleksandar

Thanks. Absolutely right in a sense that GAE comes with limitations. Maybe my question was to wide.

We don't intent to get RM working on Google but rather a developed statistical model in RM deployed as an App over the web. The App will do a PCA and Clustering (K-means). Would this be possible? How would be package and deploy this solution?

If getting RM is not a solution (cetrainly a prefered route for us) then any input on the alternatives is welcome.

Many thanks.

fastthinker

Dear Experts,
I tried to integrate RM 5.0 with GAE. Failed because java.util.ResourceBundle$Control is not in whitelist of the GAE. Has anybody met this? What is the solution? Will it work with previous version of RM, e.g.4.5?
Thank you

land

Hi,
actually RapidMiner comes with a hell of classes probably NOT on the white list. I think the most promising way is to not really do the calculations in the google app yourself but publishing an model application process as a web service using RapidAnalytics and then just docking the google app to the web service. Of course this needs an additional web server...

Greetings,
Sebastian

aleksandar

FYI we got it working for a simple example. It wasn't that hard.

THW_Mark

what is the exact advantage of using this, except for direct reporting to the web? I assume google imposes a memory limitation on the scripts that are exectured?

wessel

aleksandar wrote:

FYI we got it working for a simple example. It wasn't that hard.

Can you upload some screenshots?