Pre-requisite: Continuous Testing with DevOps
Performance Testing: It’s a Quality Assurance Methodology which ensures that Software Application / System over which the application is performing well under various workloads.
Areas on Newbies should focus on
1. Check more than status code: One might think everything is going fine when they receive the status code 200, but here’s the catch. What if the scripts are for login and it enters either bad username or password, basically it redirects you to the login page. But login page was delivered and it has the response code of 200. To avoid this, validations and assertions have to be placed on the response page, which indeed verifies the transaction were successful. A real world example could be like a performance scripts has been executed for 5000 Users and it was successful, but in reality, none of the users has been logged in. This would be a rookie mistake, Place the assertion in an intelligent way to avoid this mistake.
2. Think Time Matters: Another common performance testing mistakes are putting the improper pause and think time. Most of the common mistakes are not putting the think time or putting very less think time to 2 secs. One has to put the realistic performance scenario that emulates how the real user is interacting with the application. If the page takes 5 secs to fill the user specific information put the think time as 5. If it takes the minute to read the page put think time to a minute So, it would be better to get the realistic results instead of getting the unrealistic one.
3. BottleNeck: When you spot a bottleneck, It could get into trouble fast. It often happens that, one used to isolate the system instead of analysing root cause. One can observe this in a methodological way of doing performance test where we create the test that ramps up more slowly, so we can see the trends and collective throughput and can also see the server becoming busier. We can say the server is saturated rather than saying it’s 100% use or it’s out of memory. This might be just a symptom.
4. False Assumption: The most common are to make an assumption while running the test. The performance tool is doing its job and collecting the KPI and data. When the data is presented while running test it can look very screwed. To overcome this one should identify trends and know what’s going at the backend and should wait to complete the test results. If the failure rate is high you should stop the performance testing. Go back and do the analysis. Analysis is one of the most important aspects and most challenging part of performance testing
5. Sink Hole: Before sharing any result, I would recommend sometimes go to extreme, do not share the results until you execute the 3 runs and then you’ll able to analyse it. Most people don’t understand the performance engineering process, they want the result immediately. Most of the newbie find themselves spending the hour to analyse the anomalies, that was simply a fluke which happened once and can’t be reproduced.
6. Analysis: The way to do analysis done with most of the performance tester do.- Collecting- Aggregating- Visualising- Interpreting- Reporting