Migration Guide for TIBCO Spotfire Statistics Services Server to Spotfire Statistics Services

Spotfire Statistics Services

The various Spotfire Statistics Services can run in containers (on Linux), or simply installed as a service on a node for Spotfire Server.

Each of the services that comprise Spotfire Statistics Services is packaged as a set of Spotfire SPK packages in one SDN (Spotfire distribution file). You can add each service to the deployment area of Spotfire Server, and then deploy it to a node manager, as you would any Spotfire service. For added security, you can run the TERR engine in TERR restricted execution mode, which identifies and prohibits I/O system operations. Additionally, Spotfire script authors can trust scripts so they run for those connected to the server by a web browser.

Note: If you are currently running Spotfire data functions using the SAS or MATLAB engines, you must continue to use the Spotfire Statistics Service server configured to use the respective engine. You can vote for statistics services for these two engines in the Ideas portal:

Spotfire Server architecture with nodes running services

Important:

Different service types (for example, Spotfire Web Player and the Spotfire Statistics Services for TERR, Python, and R) use different memory models. It is likely that if you run them on the same system, you will run into memory problems. Specifically, Spotfire Web Player is tested to run multiple service instances on the same computer. The data function services are configured for all memory being dedicated to that service. A data function service does not take into account that other processes that might need RAM on the system. This means that, for production systems, you should follow these guidelines:

  • Review the system requirements for the services you intend to use.
  • A Spotfire Statistics Service supports only one instance per node manager. If you set the instance value to other than 1, then the service does not work as expected.
  • More than one Web Player service on the same node manager can improve performance on computers with large resources.
  • Different service types (for example, Spotfire Web Player and Spotfire Service for Python) running on the same node manager is not supported.
  • Only one TERR service, R service, or Python service per node manager is recommended.
  • All Spotfire Statistics Services belonging to the same Spotfire deployment area must run on node managers with the same operating system.
  • You can install a Spotfire Statistics Service on a node manager running on a computer with an operating system that is different from that of your Spotfire Server.