asbo dog I agree p2p shaping is a joke.
I have used torrents a few times in the past few weeks. Here is what I have observed.
1 - the speeds I get peak and off peak are not widly different even tho its clear my UBR is over utilised/subscribed. I am not doing anything special to evade shaping just ran utorrent on its defaults with a port specified for forwarding.
2 - when a torrent is running my tcp jitter improves, what the ...., so it seems VM's network rewards agression as I get favoured treatment for timeslots under torrent conditions.
The nighly packetloss is either DPI equipment overload, core network saturation, or peering/transit saturation, as it seems to be on every single graph my guess would be a peering issue, if its the core network it must be very close to the peering point as it affects everyone.
Here is my graph from wed 1st june the last day before VM uplifted speeds to what my UBR cant handle.
everything for the most part was fine peformance wise.
here is graph thurs day 2 june.
can see affect of uplift.
here is graph from the previous thursday when things improved as I think users got moved off my port. At 6am and 10am utilisation dropped.
Sadly I think VM realised they made a mistake as my port is the lowest utilised out of 3 in my segment, when I was told this by the CEO office I knew what may be coming and here is what did come the day after yesterday, see the utilisation jumps right up at 10am. I assume the users been moved back along with some on top to try and fix the other ports.
[img]Download Failed (1)[/img]
Do they call that network management and providing a resolution for my problem? There is a huge difference between the top and bottom graphs, in terms of the speed uplift.
Latency is worse.
Download speeds are worse.
Web browsing is slower as a result.
Downloads are more jittery as a result.
Uploads can slow down to slower than the previous non uplifted speeds yes sub 1mbit.
The only thing thats better is that bulk uploading a lot of the time is faster but everything else about the service is worse.
Also how often is a reseg actually an upgrade rather than moving users around, if there is 3 congested ports, then moving users around the 3 ports isnt going to fix the problem, ultimately either it would need lower end user speeds, less users, or a capacity upgrade. I still think now after an area has had a speed uplift there is a spending freeze on capacity hence the introduction of traffic shaping.
VM's own tests show my port is overloaded, I am not been fobbed off now so they admit is overutilised, that is something, the problem is last time it took them a year to fix it.
I would be curious for a response here from nopanic or benmgr.
Join CF
You are here: 
