# Pulp Service Performance Brian's buckets - Application Performance - Resource Allocation - Throughput # Goals 1. Understanding and scaling throughput 2. Request latency # Questions * How can we right-size the CPU and RAM limits for the api, content, and worker processes? * How do we characterize typical workloads on the api, content, and worker processes? * Do we need process recycling, and if so when? * Do we have any memory leaks? * What are the traffic volumes that we expect for the various components? * Can we horizontally scale enough to reach these traffic levels? # Workload characterization * Analyze the API calls * Analyze the task record