said by freejazz_RdJ :

Is anybody missing the fact that 14% or 16% of users using 27% and 29% of a finite resource is inequitable if the balance of users are sacrificing (by being throttled without MLPPP/VPN/SSH workarounds) their speeds in the name of equity? This kind of inequitable distribution wouldn't be tolerated with water from a common aquifer, school funding or all manner of things, why is it any more fair in this context?

said by freejazz_RdJ :

Is anybody missing the fact that 14% or 16% of users using 27% and 29% of a finite resource is inequitable if the balance of users are sacrificing (by being throttled without MLPPP/VPN/SSH workarounds) their speeds in the name of equity? This kind of inequitable distribution wouldn't be tolerated with water from a common aquifer, school funding or all manner of things, why is it any more fair in this context?

It's also clear that either throttling P2P isn't working and/or P2P isn't the major source of consumption and/or MLPPP/SSH/VPN are effective countermeasures. Perhaps the move to a per-endpoint model which would basically even these numbers out, much to the chagrin of what appears to be the high-usage client base of the wholesale customers, would be a net benefit to Bell. However, off peak, this would still allow un-even usage by wholesale clients which is fine by me... unused capacity at one point in time can't be saved for later, so make full use of it.

said by jam_bongo :

in other words the Exaflood isn't coming?

said by Guspaz :

27/14 = 1.9% bandwidth used per % user share
29/16 = 1.8% bandwidth used per % user share

During the period, wholesaler bandwidth per-user was reduced by roughly 6%

Bell's essentially misrepresenting the figures.

said by freejazz_RdJ :

Mlerner, you make points that are valid, if only on the surface in some cases.

-The increased usage potentially caused by ADSL2 tiers offered by Bell retail makes no difference. The stat is that of all the usage in the network, wholesale customer consumer nearly twice as much bandwidth in % terms than they represent in actual % of connections. There is no need to consider the sync speed in this figure, if anything, that would further prove Bell's case: "Even with our 10Mbps+ speeds, wholesale customers still use twice the bandwidth on the network than they represent by % of customers"

-The AHSSPI costs are paid by ISP's. But what does the AHSSPI include? From what I can tell, no fixed amount of capacity beyond the port at the POI. It's not like when a new AHSSPI is added there is 1Gbps of new ether distributed to all of that ISP's clients in each of the wire centres. How do would you distribute this capacity to each CO?

And by per-endpoint, I mean per DSL loop. It would be a little silly if I could circumvent a comcast-like fairness system by creating multiple PPPoE connections over 1 physical DSL line to get X time the bandwidth up to my physical DSL line's capacity. For example, I'm a big user and throttled to 1Mb/s because it's busy based on the "comcast" systems. I create 5 PPPoE sessions to allow me to get 5Mb/s. This is why it would have to be per dsl loop.

said by nigrunze :

They're "complaining" about 29%?

said by jfmezei :

December 2008, consuming 27% of the bandwidth during peak periods even though they were only 14% of the total base of end-users. As of May 2009, these numbers have increased and wholesale users are now consuming 29% of the bandwidth during peak periods even though they are only 16% of the total base of end-users. This amount is now more than double from their consumption levels prior to deployment of traffic shaping even though their share of the total base has not really changed.