I was thinking maybe that a lot of “usual” interviews, evaluations, views on testing etc do not touch the sensitive points a tester and maybe certain types of missions are needed.
Here are some I have in mind:
1) This was one I had in mind after a twitter chat with @shrinik:
Lets say Skype(doesn’t matter what provider ) added a new type of video chat that suppose to be much better regarding video and audio transmission. And they made an estimate and measurement and assume some numbers of servers for the peer-to-peer (better peer-server-peer) connections. Each of these servers have to put up with 10000 videochat conferences of some defined conditions (video stream = X data /second, audi stream =Y data /second -doesn’t matter).
The problem here is how do you test this properly, having in mind that Skype wants to know that their initial estimate, before the actual investment in hardware (budget is also limited and very important) , is accurate and reliable?
2) How would you test gmail servers from Google for 200 million users for spam filtering, so that each email sent is received in maximum 2 sec within network?
3) Start with an empty website, or an empty window application, or something that is very close to “nothing”.
And from that start giving info about the product. For example “there is nothing about this blank page, at least a help section will be need”
You can do it as an exercise, and others can play stakeholder role etc
4)Let’s say within a SMS service provider, each SMS has to be filter out by some text, and if its not filtered out the SMS is not received by the user.
And let’s assume a requirement that said in case of failover of the filtering server, a queue of up to 10 billion SMS messages have to be kept till server is back. After that number is reached, send the SMS without the filter on a FIFO (first in first out) base. The idea here is to try to save the SMS as long as possible to be filtered, but if the filter is not up, send it as it could be important for the user.
Also after the filter serve is up, the queue should be consumed within an hour and arrive to a “usual” size.
So that is what I kind of have in mind as alternative testing missions. Large scale applications are one good example. Hope others with give similar examples.ShareThis