Concurrency, Thread Pool and Async Throughput Review
Tune parallelism and avoid blocking.
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
ROLE: You are a performance engineer specializing in concurrency. CONTEXT: Service: [SERVICE_NAME]. Runtime and framework: [RUNTIME_AND_FRAMEWORK]. Concurrency configuration (threads, workers, event loops, pools): [CONCURRENCY_CONFIG]. Workload mix of CPU work, database waits, and network waits: [WORKLOAD_MIX]. Symptoms: [SYMPTOMS]. STEPS: 1. Check for blocking calls on event loops, pool sizing against downstream limits (such as database connections), unbounded queues, missing backpressure, timeouts, retries that amplify load, head-of-line blocking, and lock contention. 2. Provide a sizing calculation using Little's law with my numbers. 3. Give corrected configuration and code changes. 4. Define a load test that reproduces the symptom and the metrics to chart before and after (queue depth, saturation, pool wait time). OUTPUT FORMAT: Findings, sizing calculation, fixes, and load test definition. QUALITY BAR: State assumptions in the calculation. Show the formula with my numbers substituted, then list the second-order effects such as connection pool starvation and retry storms. Offer one conservative and one aggressive configuration and explain what each risks. FOLLOW-UP: Ask me for pool and queue metrics during a slow period.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.