When PingAccess protects an API as an OAuth Resource Server, it must validate the access token presented with each request. This validation often involves a call to the authorization server (e.g., PingFederate) for token introspection. For an API in a remote datacenter, the added latency of this validation call for every API request can significantly degrade performance. Enabling "Cache Token" on the resource configuration allows PingAccess to cache the token validation result. For subsequent requests using the same token, PingAccess uses the cached result, eliminating the need for a new introspection call and thereby reducing overall response time.

A. Reduce the Key Roll Interval on the web session: This setting pertains to the security of PingAccess web sessions, not API access token validation. Reducing the interval would increase overhead, not improve performance.

B. Move the API to a separate Virtual Host: Virtual Hosts are used to segregate configurations and listen on different host/port combinations. This does not address the network latency of validating a token or calling the remote API.

C. Reduce the number of attributes in the ID Token: The ID Token is for client-side authentication and is not typically sent to the API. The Access Token is used for API authorization, and its validation is the performance bottleneck addressed by caching.

1. Ping Identity PingAccess 7.3 Documentation, "Resources" section, "Add a resource" subsection.

Reference: "For an OAuth resource, you can enable token caching to improve performance. When token caching is enabled, PingAccess caches the token introspection result from PingFederate. Subsequent requests that use the same access token can be validated using the cached result, avoiding a call to PingFederate." This directly states that token caching is a performance-improving feature that avoids repeated validation calls.

2. Ping Identity PingAccess 7.3 Documentation, "Performance Tuning" section.

Reference: The performance tuning guide often highlights caching mechanisms as a primary method to reduce latency. It explains that minimizing external calls, such as token validation requests to PingFederate, is critical for high-throughput scenarios. Caching the access token validation is a key recommendation.