Omniverse Cache is a simple service that can be used both on users’ workstations as well as within infrastructure to optimize data transfers between Nucleus and it’s Clients ( Apps, Connectors, and others).
we recommend reviewing Nucleus Architecture for better understanding of caches.
Omniverse Cache’s idea is really simple: it can store locally, and return, files normally served by Nucleus, as long as they haven’t changed.
A few notes and details:
A Cache is something a Client connects to. This means that one Cache installation or instance will cache all requests to every Nucleus made thru it
A Cache can itself be a client of another Cache. In other words, caches can be daisy chained - meaning that one Cache can be set as an upstream for another Cache
Caches can be “chained” - meaning that one Cache can use another Cache to access Nucleusii.
This allows system administrators to design configurations of caching for their users’ workstation much more sophisticated than just a simple single cache service on those workstations.
For some practical examples of chaining, please examine notes on Caches in Infrastructure.