Saturday, December 29, 2007

Stress testing

Software engineers would be well aware of this term and might have been through this too. But, since it was my first experience and one worth writing, so I thought to share it here. This is how it went, I cannot say "this is how it goes" because I think different companies have different ways of stress testing and there actually might be better ones than what we did.

We were a team of 15 to start with. Each member in the team was provided with simplest, extremely monotonous, moronic, robotic tasks - to click on certain modules for producing results and then measure time till the results were out. Then repeat this around 150 times with different permutations and combinations. If you think we could have skipped some of them, we could not. We were asked to record time consumption from start of task to result output. No, not by looking at out watch and noting it on some notebook. But, by first installing a stopwatch on our systems and then noting down time for each of those tasks in an excel sheet with predefined rows and columns for each task and user. Since this was not enough it was informed that each task must start at same time by all team members. That would happen only when the team lead says to do so. So at 10 am when every one took their seats to become Google pigeons, confusions arose from which task to pickup first to how to report the time observed in the sheet. At last when things started this is how they continued : Team lead would say "1 2 3 start", and we would start running. Yes, running! Running, our fingers on the mouse, on keyboard - to click here and there on the product, start/stop watch, note down time consumed for producing results varying from 1.5 seconds to 15 minutes and more. Then wait till all 15 are finished with that task and move on to the next one that would be informed by the team lead.

I have heard how load testing is based on increasing load after every successful execution. But, we followed a different approach starting with 15 from 10 am to 5 pm. Cycle was repeated with only 10 users from there on till 9 pm and only a 5 user test was performed from 10 pm to 1 am. Well, that downfall in users was planned due to unavailability of manpower at night so our poor managers had no other choice.

So, at 1 am when I was finally free, my stress limits had been tested. How much stress do I feel upon doing such high level work for almost 15 hours with only tea/coffee or meal breaks and no more.

My stress testing seems to have passed as I am still writing.

P.S - Its not good to write about one's workplace because it reveals a lot about what we actually do and what people think we do. But, since it was an incident that was totally out of the blue for me so here it was to be.

3 comments:

Sandesh said...

Gettin pissed off reading it! I understand how irritated you are! But consider your lead's condition! should be even worser right? Hope this ends asap.

ankurg said...

Definitely, the lead and main testing guy who had compiled hundreds of commendable job.
There is no argument on that at all.

KbHbEjTi said...

Ankur, What am I reading here..

http://www.google.com/technology/pigeonrank.html

Is it really true?? I am damn surprised!