100M : Line quality and jitter

[qasdfdsaq]

Originally Posted by Chrysalis

That suggests to me VM perhaps put higher paying customers on lower populated channels or bump their priority, its interesting to say the least.

I thought I'd explained this already, it's probably nothing to do with channels or priority.

TCP tries to send faster and faster until it hits packet loss. When this eventually exceeds the speed someone is paying for, VM's network starts bufferbloating and then dropping packets.

People on lower speeds are more likely to hit their speed cap and therefore more likely to get their data dropped, simply from normal use.

[Chrysalis]

except jitter measurements are not TCP.

I know how TCP works thank you.

[qasdfdsaq]

Well then you'll know that TCP and ICMP packets go down the same line and when one fills it up the other will also start getting dropped.

[Chrysalis]

Your explanation is assuming the results are affected by people pegging their connections hitting speed caps and getting affected results.

The samknows test only runs when the connection is idle, if its busy then the test is skipped.

[qasdfdsaq]

No, I'm saying in any random selection of people doing a random selection of things at random times of day the ones with lower speed caps are more likely to hit speed caps.

The latency and jitter tests run continuously, it's only the speedtests that get "skipped" when the line is busy.

[Chrysalis]

on my test device they all skipped if line is busy.

[Sephiroth]

Originally Posted by qasdfdsaq

No, I'm saying in any random selection of people doing a random selection of things at random times of day the ones with lower speed caps are more likely to hit speed caps.

....

Is that right, Qasi? The 30/60/100 cap points are respectively 7/10/20 GB My spreadsheet says that (during the day):

30 meg hits the cap after 31 minutes
60 meg hits the cap after 22 minutes
100 meg hits the cap after 27 minutes

During the evening period, the cap points are halved and those times are:

L30 hits the cap after 16 minutes
XL30 hits the cap after 16 minutes
XL60 hits the cap after 11 minutes
XXL100 hits the cap after 13 minutes

Originally Posted by Sephiroth

Is that right, Qasi? The 30/60/100 cap points are respectively 7/10/20 GB My spreadsheet says that (during the day):

30 meg hits the cap after 31 minutes
60 meg hits the cap after 22 minutes
100 meg hits the cap after 27 minutes

During the evening period, the cap points are halved and those times are:

L30 hits the cap after 16 minutes
XL30 hits the cap after 16 minutes
XL60 hits the cap after 11 minutes
XXL100 hits the cap after 13 minutes

wat do u mean by that? by the cap and the mins? i am lost

[craigj2k12]

downloading at full speed on each package would trigger STM downloading for those lengths of time

[qasdfdsaq]

Originally Posted by Sephiroth

Is that right, Qasi? The 30/60/100 cap points are respectively 7/10/20 GB My spreadsheet says that (during the day):

The 30/60/100 cap points are respective 33Mbit/sec, 66Mbit/sec and 111MBit/sec.

I'm talking about the connection speed, this discussion has nothing to do with STM.

[Sephiroth]

Originally Posted by qasdfdsaq

The 30/60/100 cap points are respective 33Mbit/sec, 66Mbit/sec and 111MBit/sec.

I'm talking about the connection speed, this discussion has nothing to do with STM.

No they're not the cap points. The cap points are the data thresholds that I posted which trigger the caps.

[qasdfdsaq]

I think you're misunderstanding the term speed cap. What you're talking about is completely unrelated.