To start with, you can decide if you will deploy at your site or in the cloud:

  • On-premise deployments - SAFR Server is installed on a local machine. With this deployment model, you're responsible for configuring and maintaining SAFR Server. 
  • Cloud server deployments - SAFR Server is installed and hosted on a cloud service of your choice. With this deployment model, you're responsible for configuring and maintaining SAFR Server.
  • SAFR Cloud deployments - You use a SAFR Server that has already been installed in the SAFR Cloud. With this deployment model, RealNetworks is responsible for configuring and maintaining SAFR Server; you don't need to worry about such details. You do pay an additional fee to RealNetworks if you choose to use this deployment model, however.


Beyond that, SAFR has a wide array of options for deployment.  Following are some of the ways SAFR is highly flexible when it comes to deploying in different environments or varying requirements:

  • SAFR separates Detection and Recognition process and the ability to deploy processing notes near the cameras to reduce network traffic and recognition centrally where identities can be hosted and event to action rules can be managed
  • SAFR Servers can be deployed in a multi-level architecture because it enables following capabilities
    • 2-way identity sync - Synchronize identities based on w-directional filter conditions between multiple servers in either a hub-and-spoke or network-relay model
    • 1-way events sync to enable hierarchal relay of events from many child servers to parent server to centralize reporting and event-to-action rules regionally and/or centrally
  • Vertically and Horizontally scalable platform enabling a wide range of 
  • Ability to run single platform (detection, recognition, DB, storage, events collection and notification, etc.) at edge on devices as small as a Jetson Nano or NUC-based PC.


You may find it useful to review Large Scale Deployments for more information.