Why should CPU and memory consumption be used to identify bottlenecks?

Excessive CPU and memory consumption on critical infrastructure components, such as Delivery Controllers or Session Hosts, significantly impairs their ability to process requests for new user sessions. These components require sufficient resources to perform authentication, allocate desktop or application instances, and establish connections efficiently, and resource exhaustion will directly result in slower session creation times.

Web Interface is a legacy component, and while a data collector (also largely legacy) issues can affect brokering, its CPU/memory consumption would not directly slow down the Web Interface itself but rather the underlying session launching through it.

While a data collector's performance issues can affect services, application enumeration is primarily dependent on Delivery Controller and database performance, not solely the data collector's CPU/memory.

While high CPU/memory on a XenApp server certainly impacts performance, the local host cache (LHC) is designed for resilience during database outages, and its primary function isn't directly bottlenecked by general server CPU/memory consumption in the context of identifying typical performance issues.