Performance Testing – Traditionally Modern Analysis

The main purpose of this blog is to exhibit possible or uninhibited methods in the testing industry. Following are the few,

• Complexity Analysis
• Break-down of response times 
• Designing the architecture (Both Performance testing Environment and Environment Under Test)
• Choosing the right hardware/software (along with configurations)
• Choosing the right monitoring or Profiling tools (along with parameters)
• Choosing the timelines wisely

Note: Will cover each topic separately in the later blogs.

As soon as the performance test executions are done, we go ahead with Analysis.

As a duty of performance tester, we jot down the observations from the test executions. Following are the basic parameters considered for reporting,

Client-side statistics:

• Average Hits per second
• Average throughput
• Average/90^th percentile transaction response times and so on

Server-side Statistics(perfmon, SiteScope or any other monitoring tool):

• CPU Utilizations
• Memory Utilizations
• Network utilizations

Apart from the above if we’ve access to any other profiling tools we provide some of the following parameters,

• Heap Dump
• Thread Dump
• GC Analysis
• Drill down of high consuming methods

Usually, the above parameters were reported to the client or stakeholders.  Based on the performance testing report provided any of the following may occur,

•  Conditional Go live (Performance testers/Vendors will be happy.  If anything fishy happens these guys will not go for a bad reputation as this option was chosen by the client. Now we’re safe and Party for tonight)
• Increasing the resources (Increasing the CPU cores or Memory instances or increasing the Network bandwidth – In case if an application is not deployed on Cloud)
• Revisiting of Code