I'm using the AB command to stress test a site when-n10000-c1000 get concurrency per second around 160 when-n1000-c100 get concurrency per second around 35 when-n100-c10 get concurrency per second around 5 looking at this result a bit awesome forced, can someone analyze the real concurrency ?... I used the AB command to stress test a website.
When
-n10000 -c1000
The concurrency is about 160 per second.
When
-n1000 -c100
The number of concurrent requests per second is about 35.
When
-n100 -c10
The number of concurrent requests per second is about 5.
Looking at this result is a little awesome. Can someone analyze the real concurrency?
Or how to set the pressure test to be closer to the actual situation?
Reply content:
I used the AB command to stress test a website.
When-n10000 -c1000
The concurrency is about 160 per second.
When-n1000 -c100
The number of concurrent requests per second is about 35.
When-n100 -c10
The number of concurrent requests per second is about 5.
Looking at this result is a little awesome. Can someone analyze the real concurrency?
Or how to set the pressure test to be closer to the actual situation?
Based on your expected data, you cannot expect a server to bring you tens of thousands of concurrent requests, right?
Then, assuming that the concurrency of 100 can meet your business needs, you can set the-c parameter to 100 during stress testing.
Then, assume that you have 50 thousand requests per day, which is equivalent to PV. Then you set-n to 50000.
Therefore, the final result is close to the performance data of your real production environment, and the performance data is optimized.
After optimization, you can test the stress with a considerable number of parameters.
-CIs the number of concurrency
You can-NIncrease, for example, 10000000
Keep the stress test for a period of time and observe the pressure and load of the server.
The actual situation may be very complicated. We need to analyze the case by case.