You are viewing the RapidMiner Deployment documentation for version 9.6 - Check here for latest version
Deployment Templates for Production, for Low Latency Scoring
This template is very similar to the basic production template. This template becomes relevant when the goal is to deploy RapidMiner processes as web services where the response time of the web service needs to be really quick.
This deployment consists of the following components:
- 1 RapidMiner Server instance
- 3 RapidMiner Job Agents
- Postgres database
- 1 Real-Time Scoring Agent
- Python Environment Manager
- Real-Time Scoring Admin UI
If you plan to deploy the RapidMiner platform to a single, large (physical or virtual) machine, we recommend the template using a docker-compose based approach for the sake of simplicity.
If you plan to deploy the RapidMiner platform to multiple (physical or virtual) machines, we recommend a template using a Kubernetes based approach. We support the popular public cloud vendors' Kubernetes services, as well as your own Kubernetes cluster.
System requirements
Minimum recommended hardware configuration
RapidMiner Server
Each virtual or physical machine should at least have:
- Quad core
- 32GB RAM
- >10GB free disk space
The amount of memory needed depends heavily on the amount of data that will be processed in the Server. The RapidMiner services themselves could run even with 8Gb, however, in Production environments, we recommend 32GB or more depending on user data, in order to provide users with enough capacity to analyze data from realistic use cases.
Real-Time Scoring Agent
Each virtual or physical machine should at least have:
- Dual core
- 2GB RAM
The Real-Time Scoring Agent is a lightweight services. It has a low memory footprint and the real memory need is set by the process being run, rather than the RTSA service itself.
