Understanding QPS, TPS, Concurrency, and System Throughput Metrics

1. QPS

QPS Queries Per Second measures the number of queries a single server can handle each second, reflecting its maximum throughput capacity.

2. TPS

TPS Transactions Per Second counts completed transactions per second, where a transaction spans the request from client to server response.

3. Difference between QPS and TPS

TPS counts complete request‑response cycles, while QPS counts every individual request that may occur within a single transaction, such as multiple server calls generated by one page load.

TPS includes user requests, internal server processing, and server responses.

QPS counts each individual request, which can be multiple per transaction.

Example: a page load may generate three server requests, resulting in one TPS but three QPS.

4. Concurrency

Concurrency (concurrent degree) indicates how many requests a system can handle simultaneously, reflecting load capacity and can be derived from access logs.

5. Throughput

Throughput is the number of requests processed per unit time; both TPS and QPS are common throughput metrics.

Key throughput factors

Throughput depends on CPU consumption per request, external interfaces, I/O, etc.; higher CPU cost or slower I/O reduces throughput.

Important parameters

QPS/TPS: number of requests or transactions per second.

Concurrency: simultaneous requests the system can handle.

Response time: usually the average response time.

Relationship

QPS (or TPS) = Concurrency / Average response time.

When either concurrency or response time reaches its limit, overall throughput cannot increase further.

6. PV (Page View)

PV counts each page load or refresh, representing total page accesses.

7. UV (Unique Visitor)

UV counts distinct users visiting within a day, derived by deduplicating user identifiers.

8. DAU (Daily Active Users)

DAU measures the number of active users per day, similar to UV but focused on activity.

9. MAU (Monthly Active Users)

MAU counts distinct active users over a month.

10. System Throughput Evaluation

System design must consider CPU, I/O, external service latency, and estimate performance based on QPS, concurrency, and daily PV.

Typical approach: identify peak TPS and daily PV, relate them, and use stress testing or experience to estimate maximum throughput.

11. Basic Concepts and Formulas for Software Performance Testing

From a user perspective, the key metric is response time—the interval from an action to the result being displayed.

From an administrator perspective, important metrics include response time, resource utilization (CPU, memory, I/O), scalability, maximum concurrent users, bottleneck identification, hardware upgrades, and 24/7 availability.

From a developer perspective, considerations include architecture design, database design, code efficiency, memory usage, thread synchronization, and resource contention.