[Tams]

just for fun... here's the whole day of YMZ11


ps. no guarantee to accuracy or completeness.

[carltonp]

Originally Posted by Tams »

just for fun... here's the whole day of YMZ11


ps. no guarantee to accuracy or completeness.

Excellent. I'm so meticulous, I will go through each pause :-)

Cheers Tams

[Do Or Die]

Originally Posted by Tams »

I am not saying there is a problem, or the will be a problem.

When you first try tracert, you reported everything is ok.
Of course things are ok. Why wouldn't it?

On your above test, you reported that the esignal and IB feed are the same.
Of course they are they same, why wouldn't they?

What you have observed is of very little value.
Let me explain.


Please take a look at the tracert report again. How many hops are there?

Mine has 14 hops.
Only the first 3 hops belong to my ISP.
ie. even though I have paid premium to subscribe to a fast internet,
the extent of my influence is only 3 hops !!!
the rest of the hops belong to other ISPs along the route.
If one of them has a hiccup, you are affected.

I am not saying someone will have a hiccup.
I am not saying internet is not reliable,
and I am definitely not saying your strategy is not workable because someone somewhere during the next one hundred years will have a 100 millisecond hiccup.

What I am saying is, you have to look at all the variables,
and see if you can control/depend on those variables.

Let's go back the the observation you made in the previous post.
Of course the esignal and IB data are the same...
if they go through the same hops, they will have the same latency -- ie. same results.

Next, you have to look at your current average latency,
and then divide that into your "Pause" Strategy threshold.

ie. you have to take samples of your tracert... to see how stable is your latency.
I have a 122 ms latency right now. It might be 60 ms tomorrow, or 250 ms in the next second.
You have to find out if you have a dependable varaible.

If the pause you are working on is 1 second,
and take my example, my latency is 122 ms (I will round it up to 200 ms, because you always round up),
then I would assume my expected error will be 20%. (200 ms of 1 second)

I don't know if it would work for you, maybe, maybe not. If it is acceptable, then by all means continue with the hypothesis. Otherwise even if it is a good theory, you are limited by our current technology.

Hope it makes sense.