Part 5: Workload Models & Execution Design (Performance Testing Revision Cheat Sheet)

 

📌 Purpose

Workload modeling ensures tests mirror real-world usage. Execution design defines how the load is applied over time.

---

🎯 Workload Modeling

1. Identify User Journeys

   • Pick business-critical scenarios (e.g., login, search, checkout).

   • Assign weights based on production analytics (e.g., 70% search, 20% checkout, 10% admin).

2. Determine Load Profile

   • Concurrent Users → Number of users active at once.

   • Transactions Per Second (TPS) → Business-driven metric.

   • Think Time & Pacing → Control realism of user actions.

3. Workload Types

   • Open Model → Requests arrive at a fixed rate (good for web traffic).

   • Closed Model → Fixed number of concurrent users (good for enterprise apps).

---

🎯 Execution Design

1. Ramp-Up

   • Gradually increase load from 0 → target users.

   • Avoid sudden spikes unless testing spike behavior.

2. Steady-State (Sustained Load)

   • Maintain target load for a defined period.

   • Collect performance data under stable load conditions.

3. Ramp-Down

   • Gradually decrease users.

   • Allows system recovery monitoring.

4. Test Duration

   • Short runs → Quick validation (baseline).

   • Long runs (endurance/soak) → Detect memory leaks, resource exhaustion.

---

📊 Example Workload Pattern

• Ramp-Up: 0 → 1,000 users over 20 minutes.

• Steady-State: 1,000 users for 60 minutes.

• Ramp-Down: 1,000 → 0 users over 10 minutes.

---

🛠️ Deliverables

• Documented workload model with assumptions.

• Execution schedule with ramp-up, steady, ramp-down.

• Aligned workloads with business SLAs.