predictionio-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mars Hall <m...@heroku.com>
Subject Re: Running predictionIO on a web server other than localhost
Date Mon, 17 Oct 2016 20:31:36 GMT
Jenna,

Glad to hear you got PredictionIO running on Heroku.

The PIO engine & eventserver need to be live, reachable on the internet so that a front-end
web app can access them. Ideally all microservices (engine, evenserver, web app) are deployed
to the same cloud provider & region so that backend HTTP & database/Postgres communication
are low-latency. While it is technically feasible to have the services running in different
places, that will only degrade reliability & responsiveness of the integrated application.

*Mars

( <> .. <> )

> On Oct 17, 2016, at 12:30, Jenna Kwon <jinkwon06@gmail.com> wrote:
> 
> I tried the github guide out and was able to deploy the event server & classification
engine to heroku. I have my locally running webapp communicate with them via HTTP. Thank you!
> 
> I have a remaining question. Must the eventserver & predictionIO engine be deployed
on the cloud if I want my webapp running on a web server (instead of on localhost) to communicate
with them via HTTP? 
> 
> Best Regards,
> Jenna Kwon 
> 
> On Tue, Oct 4, 2016 at 11:28 AM, Mars Hall <mars@heroku.com> wrote:
> Hi Jenna,
> 
> Today at Dreamforce (the Salesforce conference) we're presenting this PredictionIO Heroku
buildpack in a developer session called "Exploring Machine Learning on Heroku":
> 
> https://github.com/heroku/predictionio-buildpack
> 
> The README is a tailored guide to getting an attribute-based classifier running. To use
the buildpack for engines of your own choosing, follow the more generalized (and more complex)
CUSTOM docs:
> 
> https://github.com/heroku/predictionio-buildpack/blob/master/CUSTOM.md
> 
> If you decide to try it out, please post Github issues with questions, problems, or feedback!
> 
> *Mars Hall
> Customer Facing Architect
> Salesforce App Cloud / Heroku
> San Francisco, California
> 
> 
> On Tue, Oct 4, 2016 at 05:34 Sunny <sunnystarin@gmail.com> wrote:
> Hello Jenna,
> 
> Deployment in local is fine if you are running lightweight jobs and using it just for
hands-on and learning purpose, no need to bear extra cost of the additional computing power
of cloud services.
> In case you are into heavy weight research work or running a production system and wants
it to be accessible from around the globe, private or hybrid or using available cloud services
like AWS is must.
> 
> -- Tapo
> 
> On Tue, 4 Oct 2016 at 06:23 Jenna Kwon <jinkwon06@gmail.com> wrote:
> Dear Mr. Szeto,
> 
> Thank you so much for your reply. I saw some posts demonstrating deployment of event
server & engine via Heroku. I also saw that there used to be deployment through AWS. In
what case would I want to do that instead of keeping them local?
> 
> Best Regards,
> Jenna Kwon 
> 
> On Mon, Oct 3, 2016 at 5:45 PM, Donald Szeto <donald@apache.org> wrote:
> Hi Jenna,
> 
> You probably want to have a reverse proxy in front of your app (localhost:3000) so that
it is exposed to the Internet via standard ports (80 for HTTP or 443 for HTTPS). Your app
can communicate to the event server and engine over localhost. They don't need to be exposed
to the Internet. Your app should be using one of PredictionIO SDKs to interact with the event
server and engine.
> 
> Regards,
> Donald
> 
> On Sun, Oct 2, 2016 at 3:07 PM, Jenna Kwon <jinkwon06@gmail.com> wrote:
> Hello!
> 
> I am a beginner so please excuse me if my question is rudimentary.
> I've followed a demo for Tapster Comics Recommendation (http://predictionio.incubator.apache.org/demo/tapster/)
and have a good understanding of deploying the engine, app and event server on localhost.
> 
> My app ran on localhost:3000, engine ran on localhost:8000, and eventserver ran on localhost:7000.
> 
> I would like to now deploy these on a web server. The "app" would be a specific website.
How would I hook up and deploy eventserver and predictionIO engine to gather information from
that specific website? 
> 
> Would it just be a matter of changing environmental variables in pio-env.sh (in predictionIO
installation directory) as well as those in app's directory? For example, would I substitute
"localhost" with a web server? 
> 
> Thank you!
> 
> Best Regards,
> Jenna Kwon 
> 
> 
> -- 
> Jenna Kwon
> B.S. Computer Science '17| Georgia Institute of Technology 
> Denning Technology and Management Scholar 
> GT Tennis Club | GT Trailblazers
> www.jennakwon.com
> (541)908-2963
> 
> 
> 
> 
> -- 
> Jenna Kwon
> B.S. Computer Science '17| Georgia Institute of Technology 
> Denning Technology and Management Scholar 
> GT Tennis Club | GT Trailblazers
> www.jennakwon.com
> (541)908-2963
> 
> 
> 
> 
> 
> 
> -- 
> Jenna Kwon
> B.S. Computer Science '17| Georgia Institute of Technology 
> Denning Technology and Management Scholar 
> GT Tennis Club | GT Trailblazers
> www.jennakwon.com
> (541)908-2963


Mime
View raw message