Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Deployment of RapidMiner on AWS/ Azure Cloud

RG2000

I am exploring using either the AWS or Azure Cloud instances. I wanted to confirm on the following
a) Access to publishing from studio to AWS or Azure Cloud is not available for free studio
b) Are there restrictions on number of models that can be hosted- any linkage between number of models and server size
c) Are these models published on 'hot mode'- or on 'cold mode'- to be brought up when called
d) wanted to understand security capabilities, logs
Where would documentation for cloud based services be available?

Find more posts tagged with

AI Studio

AWS Azure

Accepted answers

sgenzer

hi @RG2000 I'm sorry for the delay but I was away last week. Let me see if I can answer some things here...

(a) I don't see any reason why a local installation of RM Studio (free edition) cannot connect to a cloud install of RM Server. I just tested connecting a free local RM Studio license with one of our AWS RM Servers and it works fine:

Image: https://us.v-cdn.net/6030995/uploads/editor/c1/f973rlxtqa7i.png

(b) there is no limit per se to the # of models on one server. Obviously you need to size your server to what kind of models you're running. You can probably run 100 Titanic Naïve Bayes models on one server every minute if you want, whereas running one 1.5M row x 100 column data set with DL or GBT is going to eat up some serious resources.

(c) you can have models built on a reoccurring schedule (CRON jobs) or on-demand as an API call.

(d)
https://docs.rapidminer.com/latest/server/configure/security/
https://docs.rapidminer.com/latest/server/configure/logs/

Scott

All comments

rfuentealba

Hello @RG2000,

This message is for @sgenzer or @leti probably. Asking them for their support here. I can only answer b) and possibly c):

b) server size doesn't have anything to do with the number of models, but it does have to do with the size of your data, probably.
c) both modes are supported, as far as I know.

All the best,

Rodrigo.

RG2000

@leti, @sgenzer

Could you share details around (c) and (d). Any documentation would be great

leti

All of our server documentation fro Azure and AWS can be found here https://docs.rapidminer.com/latest/server/installation/cloud_images.html and security documentation can be found here https://docs.rapidminer.com/7.6/server/administration/security/

sgenzer

RG2000

Thanks- just to be extremely clear, is the server capacity driven by the number of active APIs being called, and not the number of APIs deployed (i.e. I can host a large number of models on the server, as long as concurrent usage is low)

Nikouy

Do you know if it's somehow possible to scale up your processing power/rapidminer fleet based on the number of API calls?

sgenzer

hi @Nikouy do you mean automatic scaling such as an AWS Auto Scaling kind of thing?

MartinLiebig

@Nikouy,

you would likely use a RM Real Time Scoring Engine in Kubernetes for it.

Cheers,

Martin

Nikouy

@sgenzer

Yes exactly, such as EC2 autoscaling groups or auto scaling dockers in AWS ECS.

@mschmitz Do Kubernetes clusters only work with RM real time scoring, for which I need to have previously developed and trained model in Rapidminer studio?

Do you provide educational licenses for RM real time scoring?

Are these the images available for RM Server too? https://docs.rapidminer.com/latest/server/deploy/kubernetes/ ?

Thanks