GPII Deployment Structures

From wiki.gpii
Jump to: navigation, search

Summary

This document provides GPII deployment structures for three use cases:

  • Public work stations
  • Personal devices
  • Anonymous/non-cloud use case

Note:

  • Each double-line box in diagrams represents an individual node instance
  • The components that are not in double-line boxes, user listeners and environment reporter in these cases, may or may not be node instances
  • The communication among all instances uses HTTP requests

Public work stations

GPII deployment - hybrid.png

In this deployment, the user token and device information are sent from the local machine to the cloud. The matchmaking happens in the cloud and the settings are filtered before being returned to the local machine. This is a typical use case for public work stations such as American job centers.

Personal devices

GPII deployment - cloud.png

In this deployment, only the user token is sent from the local machine to the cloud. The user preferences and solutions registry are returned by the cloud and the matchmaking happens at the local machine. This is typical for personal devices where users don't want to leak their device information and are fine with their preferences being sent back and used on their own devices.

Anonymous/non-cloud use case

GPII deployment - local.png

In this deployment, nothing is sent from the local machine to the cloud, everything performs locally. This is the use case called "Bearer token" in which the user token is the encoded user preferences. This allows people to use GPII without identifying themselves.