使用JMeter + Badboy 测试基于事务粒度的性能测试

 The following 4 transaction based performance result is done by JMeter+Badboy tool.

 

Support the scenario is as follows:

Guest view products in estore -> Guest Add item to shopping cart -> Guest check out -> Guest register his/her account->Guest login with his/her account -> Guest choose shipping method->Guest set shipping address->Guest set payment method->Guest finish checkout->System generate order no.

 

We can use badboy to record the script to imitate this scenario and export to JMeter format ,and JMeter can use this script to run performance test.

 

 

TC1->Ecshop Guest Checkout Transaction (100 users /500 transactions)

 

TC2->Ecshop Guest Checkout Transaction (100 users/5000 transactions)

 

TC3->Magento Guest Checkout Transaction (100 users /500 transactions)

 

TC4->Magento Guest Checkout Transaction(100users/ 5000 transactions)

 

 

 

The following is the conclusion:

 

(1)    Ecshop can support 100 users/ 5000 transaction very well, because Error%=0

(2)    Magento can’t support 100 users/5000 transactions ,even can’t support 100 users/500 transactions ,Error% in average is larger than 60%

(3)    Login Page/Item Detail Page/Shopping Cart Page ,the request to these three pages in the whole transactions has low pressure ,because their Error% is less than 2%.

(4)    Begin Checkout /End Checkout page has medium pressure ,because their Error% is 17%, neither too high nor too low.

(5)    Shipping Method /Shipping Address/Billing Method has high pressure ,because their Error% is larger than 98%

(6)    To this Transaction-related cases ,500 transactions/5000 transactions has almost the same pressure ,because system can utilize the load (62.79% vs 61.07%)

 

Best Regards,

 

Charles (Canton team)

Senior Software Engineer