I am trying to understand what type of an Amazon AWS node I should get to install KNIME Server on. The AWS Server would run 10 licenses of KNIME Server, KNIME Big Data, and KNIME Spark. I would like to know the minimum recomendation and the preferred recommendation.
This is what I would like to know:
- What type of Amazon node? M4, R4, C4, ….?
- How much memory on the node?
- How many cores on the node? I know that KNIME imposes a limitation of 16 cores for 10 licenses, but having said that, how many cores should the machine have?
- How much disk space?
Perhaps others have different opinions, but here is my initial recommendation. Of course the workflow workload can vary highly between deployments so there is no correct answer. You can get a reasonable idea from benchmarking prototypical workflows and estimating the number of concurrent users.
The following would be the highest specced instances with 16 cores.
m4.4xlarge 16 cores 64Gb RAM
c4.4xlarge 16 cores 30Gb RAM
r4.4xlarge 16 cores 122Gb RAM
r4 is probably the most appropriate (RAM is usually limiting).
Disk space fairly easy to scale with EBS. Choose 1TB to start with (unless working with massive datasets).