Jmeter:- How to increase and Decrease and then increase decrease TPS using jmeter?

Question

I have 300 VUsers (Threads) and ramp up 600 seconds and total number of requests (http Request) 5 , Response time for each request is 2 seconds so total 10 sec will takes to complete one loop,

I was using 2.13 Jmeter Version but now moved to Jmeter 4.0 Version.

App Server : Linux

Jmeter Placed at Windows Server which highly configured and High Bandwidth and placed in USA.

Scenario like given Below.

Login Application 
Create Order 
Active Order 
Disconnect Order 
Cancel Order 

I want to Achieve TPS like given below

1 TPS to 10 TPS for 60 Minutes 
2 TPS to 4 TPS for next 60 Minutes 
2 TPS to 8 TPS for next 60 Minutes  
... and so on

I have used "jp@gc - Throughput Shaping Timer" But unable to achieve expected TPS.

Start RPS    End RPS    Duration

 1           10         3600

 2           4          3600

 2           8          3600

• Total Number of Users- 300
• Ramp Up 600 Seconds
• Duration of Test 10800 Seconds
• Targeted TPS 1-10 TPS
• Achieved TPS 2 TPS

Please suggest what should I do to achieve such Fluctuated TPS

Please suggest the strategy need to follow.

Which Plug in I can use and how to calculate TPS in it.

Answer 1

It might be the case your application simply cannot respond as fast as you're expecting it to do, i.e. if you're able to achieve only 2 TPS it means that application cannot process more than 20 requests per second.

So I would recommend looking on Linux host instead of trying to manipulate JMeter settings, the reasons could be in:

• your Linux server simply lacks resources (CPU, RAM, Network, etc.) so make sure to monitor the above metrics using i.e. JMeter PerfMon Plugin
• your application infrastructure configuration isn't suitable for high loads, i.e. application server and/or database are running on developer default settings in terms of thread pools or allocated memory
• your application is using inefficient algorithms or operates large data objects, try re-running your load test and enable profiler tool telemetry to detect where your application is spending the most time, it could be caused i.e. by recursive long function call or slow DB query