High Availability Setup - Load Balancer

RaviSwamyAptus

Hello,
The RM documentation on Load Balancer say "The load balancer should be a dedicated machine which is only responsible for redirecting traffic and load balancing several RapidMiner Server instances."
Is there any guideline on the system requirements/configuration for Load balancer ?

sgenzer

cc @mmichel @ZSzatmari

ZSzatmari

Hi @RaviSwamyAptus,

it is a good architectural pattern to keep the LB independent from the backend servers. I won't say, that it should be a dedicated server, but should be independent from the RapidMiner Server boxes.

I propose to use HAProxy as a LB implementation, because it can be easily configured and also provides session stickiness options. It doesn't require too much resources. You can be good enough with 2 CPU cores and 2-4 Gb memory based on the connection requests per second (details: https://www.haproxy.com/documentation/hapee/1-5r2/getting-started/os-hardware/)

For the configuration-relaed question I propose to check our docs page (https://docs.rapidminer.com/latest/server/high-availability/loadbalancer.html), we have there a good example configuration to start with.

sgenzer

cc @mmichel @ZSzatmari

ZSzatmari

Hi @RaviSwamyAptus,

it is a good architectural pattern to keep the LB independent from the backend servers. I won't say, that it should be a dedicated server, but should be independent from the RapidMiner Server boxes.

I propose to use HAProxy as a LB implementation, because it can be easily configured and also provides session stickiness options. It doesn't require too much resources. You can be good enough with 2 CPU cores and 2-4 Gb memory based on the connection requests per second (details: https://www.haproxy.com/documentation/hapee/1-5r2/getting-started/os-hardware/)

For the configuration-relaed question I propose to check our docs page (https://docs.rapidminer.com/latest/server/high-availability/loadbalancer.html), we have there a good example configuration to start with.