This document outlines a top-down method for analyzing performance issues in a multi-tier architecture. It describes starting with the user scenario, measuring end-user response times, and setting a target response time. It provides an example infrastructure with user, client, server, and database tiers. The method involves understanding what each component requests from the next one to identify where a performance issue lies - whether with the user, client, server, or database. By drawing the full architecture map and working top-down, the specific source of a performance problem can be found.