NNSquad - Network Neutrality Squad
[ NNSquad ] Re: As predicted: TheBitTorrent vs. "traffic shaping" arms race
And around we go. I thought I already suggested something like this. The ISP's can't know what individual customers' preferences are. Therefore, the customer must be allowed to mark their traffic as they see fit. However, some customers would undoubtedly mark all of their traffic as high-priority, "stealing" high-priority bandwidth from other users. Therefore, let the customers mark their own packets, but have the ISP enforce to the published SLA's as far as how much traffic of different priorities is allowed. A low bandwidth cap for VoIP traffic should cover a single call using most encodings, so the ISP should remark, or drop, traffic coming from individual customers that is above this rate. Customers should have the option of marking traffic as scavenger, meaning it is the preferred traffic to be dropped if any traffic at all has to be dropped or shaped (delayed in the outbound queues on devices). BT traffic, FTP downloads, and other similar large but not necessarily time-dependent traffic could fall into this category. "Normal" traffic, such as HTTP traffic, is not treated as high-priority, neither is it the first to be dropped / shaped. ISP's can offer to prioritize customer traffic for them, but the customers should retain the right to send traffic with their own chosen DSCP values, and ISP's should be barred from modifying those values if they fall within the published rate limits. I don't agree with the statement "it wasn't that long ago when we were all blithely assuming that VoIP and such things would simply be solved with QoS bits. You'd think we'd learn that data may not have desires and motives, but the people sending and receiving it do." QoS IS the only way to "solve" the traffic behavior requirements of VoIP. QoS has been used over and over and over again in enterprise networks in order to guarantee VoIP quality. The issue with VoIP over the Internet, as opposed to within an enterprise network controlled by a single entity, is the multiple ownership of the internetwork. By its very definition the Inter-net is composed of various networks owned by various companies that are tied together. Because there is no overriding agreement between ALL network members as far as how to "trust" the DSCP markings of packets there is a perception that QoS won't scale. I don't believe that is true. If all of the ISP's agreed or are forced to enforce a common service policy, to prevent a few users at a single ISP from flooding the Internet with high priority traffic, then it just could work. Perhaps I do agree with part of that statement. It won't be "simply" solved, but it can be solved with the proper configuration and agreements between network owners, or regulations. Because of the reluctance to make any progress on this type of solution, precisely because it would commoditize the Internet and make it harder for ISP's to "monetize" their investment by offering their own services, regulation is likely required. With an Internet that was fully implemented in this model ANYONE would be able to offer VoIP or other services and be able to guarantee service quality by riding on the guarantee that ISP's would give as far as how much bandwidth were available for high priority traffic. That's something that is clearly not in the interest of the ISP's. Just like there are regulations as far as how much you can charge and what service levels you must provide if you go into business and provide transport for humans in your personally owned car (a taxi business) there are likely regulations needed to ensure minimum service levels and rates for people / businesses that use their own personally owned network to provide transport for packets. Fred Reimer -----Original Message----- From: nnsquad-bounces+freimer=ctiusa.com@nnsquad.org [mailto:nnsquad-bounces+freimer=ctiusa.com@nnsquad.org] On Behalf Of Wes Felter Sent: Wednesday, February 20, 2008 4:25 PM To: Kee Hinckley Cc: nnsquad@nnsquad.org Subject: [ NNSquad ] Re: As predicted: TheBitTorrent vs. "traffic shaping" arms race Kee Hinckley wrote: > Based on those assumptions, I see three main issues to address. > c) When services *can* get priority. > You may want to download that movie to your TiVo right now, but if your > neighbors are using VoIP, you're just going to have to wait. Only the > ISP can see the overall picture. I am not sure about this part. Obviously my high-priority traffic should have priority over my low-priority traffic, but why should my neighbor's traffic (of any type) have priority over mine? I would suggest that during times of congestion each customer should get a fair share of bandwidth and then you should do prioritization within each of those shares. This gives customers no incentive to cheat by marking all their traffic as high-priority, nor does it penalize customers who send no high-priority traffic. (As you may tell, I tend to think of this from an implementation perspective.) Wes Felter - wesley@felter.org
Attachment:
smime.p7s
Description: S/MIME cryptographic signature