View Single Post
Old 01-12-2015, 11:35   #35
ianch99
cf.mega poster
 
ianch99's Avatar
 
Join Date: Feb 2012
Posts: 4,725
ianch99 is cast in bronzeianch99 is cast in bronzeianch99 is cast in bronzeianch99 is cast in bronze
ianch99 is cast in bronzeianch99 is cast in bronzeianch99 is cast in bronzeianch99 is cast in bronze
Re: Virgin Media’s New 200Mbps Package Drops to 130Mbps at Peak Times

Quote:
Originally Posted by horseman View Post
SK certainly have the graphs but ultimately it requires VM(or any ISP if we want a comparison) to authorise their release.
However you haven't clarified whether you literally want UK monthly average "peak downstream rates" and/or "peak upstream rates" or perhaps download speeds during peak hours for weekdays during calendar month, by week or even days?

Many (undisclosed) stats I can blame VM for but I can't blame them for not publishing until all their ISP competitors agree to do likewise!

As Disraeli once said: "There are lies, damn lies and then there are statistics"
As already stated/implied by others 130Mps is only an "average" of 3 SK "normalised" hourly plots of 8,9 & 10pm for the percentage of 700 to 1400 SK trialists contributions UK wide on V200 tier for a October calendar month (of weekdays) which represents 65% of headline speed.
However Sept yielded 107 for 152 Tier = 70% and you find lower tiers varied between 83% for 100Mbps and 96% for 50Mbps.

All that illustrates is what some of us already know and most already suspect that there's a percentage of cable segments in congestion and thus higher tiers typically suffer the worst percentage degradation on headline during peak hours.

That doesn't mean VM is not addressing "congestion" in ALL areas. If I compare my V200 SK D/S speeds for
October = 118(8-10pm) 173 daily avg
November = 156(8-10pm) 189 daily avg

Whereas using a SK monitor on another Brighton node/cable segment:
Oct = 124(8-10pm) 194 daily avg.
Nov = 180(8-10pm) 201 daily avg

Neither of these indicate which of the two cables segments had an Arris E6000 and 16channel upgrade on 15th October.... or whether the Hubs were on same 8 channel or 16channel bonding group for the whole period?

VM (nor SK) publish how many samples (SK Trialists) exist but from OfCom figures the active samples (for all BB ISP's) appear to be ~1400.
My own unauthoritative guess is only a max of 700-1000 samples are included at any one time and because we also don't have the average Tier demographics either I personally suspect these are in a 60:30:10 ratio which yields max 100 Vivid200 across UK.

Thus I have no way of formulating how the remaining 98 on V200 Tier compare nor how quickly upgrades/speedboosts will rollout but as crude dipstick to meet 90% rollout (VM advertisement) of V200 by year end then I'd expect the Oct 130 peak will need to raise to 143 in Nov, 156 in Dec and 169 there on.

As stated I'm not sure what "peak speed" profile you're seeking and for all the aforementioned reasons I'm not sure exactly what conclusion you'd draw from it even if VM unilaterally provided it?
However fwiw here's hour-of-day average for one BN3 node in November:
https://www.dropbox.com/s/egs14jbhtx...38.28.png?dl=0
This is really interesting, thank you. I guess the metric that I would be interested in, as it affects us directly, is what percentage of segments are overutilised. Now how you define "overutilised" could be subjective of course but maybe a crude definition is how many segments have peak speeds that drop below the headline speed of the tier beneath. Or, you could use the definition of segments have open utilisation related fault tickets ..

VM would never release this numbers but if you look at the VM Speed forum (http://community.virginmedia.com/t5/Speed/bd-p/Speed), it seems a lot more active with possible utilisation related complaints than when I was last frequenting the place ..
__________________
Unifi UCG Ultra + Unifi APs | VM 1Gbps
ianch99 is offline   Reply With Quote