?

Log in

No account? Create an account

Preliminary results of my ping experiment - Journal of Omnifarious

Dec. 16th, 2008

06:43 pm - Preliminary results of my ping experiment

Previous Entry Share Next Entry

I've been running tests of ping times and the results look like I'm right. Ping times do seem to fit an exponential probability distribution, which also means that the facile statistical analysis that mtr and ping now do are misleading in terms of predicting what future ping times are likely to be or in saying something about how much 'jitter' a network path really has.

Current Location: 47.6231, -122.35539
Current Mood: [mood icon] pleased

Comments:

From:niac
Date:December 17th, 2008 06:26 am (UTC)
(Link)
(Reply) (Thread)
[User Picture]
From:omnifarious
Date:December 17th, 2008 05:48 pm (UTC)
(Link)

That's true. I think that affects the validity of the precise numbers, but I suspect it doesn't really affect the validity of the overall probability density function describing the data.


OTOH, I really should run some ping tests involving TCP. It's a pity the 'port unreachable' reply comes back via ICMP. This means I'll likely have to have cooperating computers on either end. Or, maybe I could make an SMTP client that waits for the server response header then sends 'QUIT'.

(Reply) (Parent) (Thread)