Multi-node workers setup requirements

Mahadeva

Hi All,

I'm exploring the Alteryx scaling features via multi-node architecture based on the workflow load etc. I would like to understand the requirements, pre-requisites if any for setting up multiple server instances to act as worker nodes.

1. Assuming we have 1 server instance acting as controller, and multiple server instances acting as worker nodes, what kind of Alteryx license should be used in these controller and worker instances, is it Alteryx Server license or Alteryx Designer license.

2. If we have to use the multi-node architecture for scaling, is it mandatory to publish/host all the workflows/Apps that are developed into the company's Alteryx Private Gallery ?

3. Can we use multi-node architecture for scaling, without hosting our apps in Alteryx Gallery. i.e. to use this feature on the apps that are maintained locally within the Alteryx designer and which are triggered, scheduled within Alteryx scheduler. If so how do we share the load across multiple worker nodes ?

4. Should all the apps/workflows be hosted only in the server instance acting as controller and nothing on worker nodes?

Appreciate your response and support. Thanks in advance.

fmvizcaino

Hi @Mahadeva ,

Here are the answers:

1. Assuming we have 1 server instance acting as controller, and multiple server instances acting as worker nodes, what kind of Alteryx license should be used in these controller and worker instances, is it Alteryx Server license or Alteryx Designer license.
   1. It is called a 2-core add-on. Each 2-core add-on costs 1/4 of the Alteryx Server license. You must have at least a 4-core add-on on a worker node to be compliant to Alteryx suggestion.
2. If we have to use the multi-node architecture for scaling, is it mandatory to publish/host all the workflows/Apps that are developed into the company's Alteryx Private Gallery ?
   1. It is not mandatory, but advisable. With Alteryx Server, you have version control, sharing capabilities, scheduling options and all the governance and control of your assets.
3. Can we use multi-node architecture for scaling, without hosting our apps in Alteryx Gallery. i.e. to use this feature on the apps that are maintained locally within the Alteryx designer and which are triggered, scheduled within Alteryx scheduler. If so how do we share the load across multiple worker nodes ?
   1. If you have workflows published to the Gallery, the controller will be responsible for managing the queue and distributing it between the worker nodes.
   2. With the SchedulerDB, one of the options will run the workflow locally and the other will be run at the controller node directly (this I would ask Alteryx support)
4. Should all the apps/workflows be hosted only in the server instance acting as controller and nothing on worker nodes?
   1. Workflows published to Gallery will be inside MongoDB. Usually at the same server as your controller node
   2. With the schedulerDB you have two options:
      1. Workflows running locally, meaning that you are connected to the controller node, but the machine responsible for running the workflows is the one your workflows are located. It could be the controller node or any worker node
      2. Workflows running on schedulerDB. You are saving a copy of the workflow on the SchedulerDB and the controller node will be responsible for deciding where the workflow will actually run.

I highly suggest using the Gallery to host all of your workflows. You will have the same features you have in Designer and so much more! I`m happy to schedule a call and walk you through a server demo if you have never seen it and we can discuss the features.

Best regards,

Fernando Vizcaino