Before getting to the actual question I have, allow me to give a bit of context. I'm developing for my portfolio a series of Python scripts to probe a server (that is, to know the status of the connection). I won't get into the beef of the code, but a question has arisen and I'd like to know if you would be able to give me a hand with it.
I know that the greatest sign of an unstable connection is notable response speed differences along a huge amount of packages (this last condition being just the statistical law of "the greater the sample, the closer to the full population"), and a friend of mine who is specialized on this has recommended me to do streaks of at least 50 packages for this.
I have implemented in-code the statistical functions needed for this task: the average and the standard deviation (on a sample), which allow me to get the values which the response time moves inbetween.
Now, I know I said the word "notable" before, yet at this lays our problem. As you can guess I'm a newbie when it comes to networking (I decided to take on this project as to have a bit of diversity and not just focusing on a single thing), and I know barely the bare minimum.
Plus, I am not able to work that well when it comes to small quantities: were this happening on the 70's when response times were long as hell there wouldn't be that much of a problem but now it's measured on milliseconds, which is too much of a short time for me to comprehend notable differences.
That's why I'm asking for an informed opinion on the matter: how much undulation should I take in account to consider a connection unstable? Thanks beforehand and excuse the inconvenience
No comments:
Post a Comment