NNSquad - Network Neutrality Squad
NNSquad Home Page
[ NNSquad ] Re: Is network unneutrality necessarily bad?
Hi, One problem with the situation here is that people are competing for bandwidth in the first place. When I buy a 3mbps connection from Bell, Bell should give me a 3mbps connection and just charge the right price for it. I should not be competing with my neighbor: I should have 3mbps and that's all. Now *I*, on my own router or connection, can "unneutralize" whatever I want and prioritize VoIP traffic or SSH connections (which I do, btw) but Bell shouldn't be allowed to do that. Why? Because they are in a conflict of interest. Their interest is not to provide good QoS to everyone, it's to maximize their return on investment. So what ISPs usually do is under-provisionning: say they sell 1000 3mbps ADSL connections, they're *not* going to buy an associated 3gbps pipe on the other site. They'll buy less bandwidth and hope that the users don't go all together download stuff at the same time. In a way, that is justified if the nominal bandwidth usage is way below that number, but that's evidently not the case: if people use all their available bandwidth, not only they get charged extra for going over their bandwidth caps (in the case of Bell and Videotron, for example), but they also end up degrading the overall network quality. Not because they are *abusing* the network, they're using what they bought. It's because the network is not capable of answering the load, because Bell wanted to make a buck. So to answer your questions more clearly: Yes, network unneutrality is necessarly bad, unless it's sold as such. On Thu, Nov 08, 2007 at 09:57:15AM -0800, Nick Weaver wrote: > So why should (if I'm willing to pay) this traffic not receive better > than best effort service. Because it shouldn't be necessary, in theory. Where I come from, in a datacenter, when you buy a 100mbps pipe, you get 100mbps, and that's it. You're fully entitled to max out that pipe (and you pay for it too!) and it would be considered downright fraudulent to shape that connection based on the type of traffic you have (for example). I reverse the question: why should it be different for the last mile? > Likewise, why shouldn't an ISP give Bittorrent and similar > bulk-transfer products worse-than-best-effort? These applications are > obviously noninteractive, high data rate transfers, so why should my > SSH connection have to compete with my neighbor downloading a big DVD? Exactly: there's no valid reason why you should have to compete with your neighbor. The solution to the problem, however, is not to discriminate on some traffic. For me, that is fraud, pure and simple. The solution is to provide the adequate bandwidth to the consumer. If you pretend you sell 3mbps, provide 3mbps, and charge for it, that's all. > How does one draw the line between legitimate traffic shaping and > outright abuses (eg, FIN-flooding seeders, HTTP add-injection, > penalizing a competitor's VOIP service since the network operator is > also the telephone company)? In my opinion, there's no legitimate "traffic shaping" outside the one that will ensure the consumer uses the bandwidth he paid for: if the consumer pays for 1mbps even though the ADSL modem syncs at 3mbps, then shaping the whole pipe at 1mbps is appropriate. Any other provision is a break from network neutrality that should be explicitely stated as being part of the service. Let's face it: people buy "ultra high speed" gizmos to download media off the internet, not for VoIP, not for SSH. If the ISPs can't provide the bandwidth the consumers are paying for, I repeat it, it's fraud. A. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD4DBQFHM4FtWGBzs0AjcC8RAqUwAJ4pqpCm/Fp4pPiO1PgWIcrXEwtEJACYkbCd o5iMDtHewNpdDkHBcFn6ww== =5+Rn -----END PGP SIGNATURE-----
_______________________________________________ NNSquad mailing list information: http://lists.nnsquad.org/mailman/listinfo/nnsquad org