3

I am using JMeter for testing my REST API. The API must handle around 100 tps load. I am using following hardware for generating load:

CPU - 8 Cores
RAM - 16 Gb
3.0 GHz Intel Xeon Platinum Processor
_JAVA_OPTIONS = -Xms1g -Xmx12g

Now I have single http GET request. My JMeter configuration:

number of users - 10000
ramp up time -  100 seconds

The above configuration means 100 concurrent users per second, and that one request is equal to 1 trasaction per second. Am I interpreting it correctly? Also, how should I plan to test my API for 500tps load? My server runs out of memory if I increase the number of threads.

Super Jade
  • 303
  • 2
  • 5
  • 17
user978601
  • 31
  • 1
  • 1
  • 2

1 Answers1

2

The above configuration doesn't mean 100 concurrent users per second. It also does not mean one request is equal to 1 trasaction per second whatever you mean by this.

The above configuration means that

  • JMeter will start 100 users each second
  • The users will start execute the request as fast as they can
  • Once virtual user finishes the request it will be shut down

The actual number of concurrent users and transactions per seconds depends on your server response time, i.e. if response time will be 1 second - you will have 100 transactions per second, if response time will be 2 seconds - you will have 50 transactions per second, if response time will be 0.5 seconds - you will have 200 transactions per second.

If your goal is to achieve 100 or 500 tps - I would recommend using Throughput Shaping Timer and Concurrency Thread Group combination. They can be put together using feedback function so JMeter will automatically start new threads in order to reach/maintain the defined concurrency.

Both test elements can be installed using JMeter Plugins Manager

Dmitri T
  • 596