Comments on: Smackdown! http://bennett.com/blog/2006/06/blind-men-and-the-elephant/ A regular old blog Fri, 04 Sep 2009 23:51:11 +0000 http://wordpress.org/?v=2.9.1 hourly 1 By: Richard http://bennett.com/blog/2006/06/blind-men-and-the-elephant/comment-page-1/#comment-296386 Richard Fri, 09 Jun 2006 21:19:26 +0000 http://bennett.com/blog/?p=3918#comment-296386 I'm not opposed to ISPs making contractual agreements to provide QoS-enabled services to anybody who wants to buy one, I'm just saying that traffic classification based on stream contents is an inexact science. NSPs already provide different service levels to their various customers and have for a long time. This is part of today's "neutral" Internet. I think an ISP has the right to handle DiffServ markers any way they want, because before you use DiffServ you're supposed to do some specifying of the stream. If you're requesting a service level you haven't contracted for, you don't get to use it unless the other guy pays for it. The only issue I have with this is a truth-in-advertising one. If you've paid for QoS and requested it appropriately, then you better get it. If you haven't paid for it and they give it to you anyway, that's their choice but they aren't required. But I don't have a problem with tiered service agreements, you see. I’m not opposed to ISPs making contractual agreements to provide QoS-enabled services to anybody who wants to buy one, I’m just saying that traffic classification based on stream contents is an inexact science. NSPs already provide different service levels to their various customers and have for a long time. This is part of today’s “neutral” Internet.

I think an ISP has the right to handle DiffServ markers any way they want, because before you use DiffServ you’re supposed to do some specifying of the stream. If you’re requesting a service level you haven’t contracted for, you don’t get to use it unless the other guy pays for it. The only issue I have with this is a truth-in-advertising one. If you’ve paid for QoS and requested it appropriately, then you better get it. If you haven’t paid for it and they give it to you anyway, that’s their choice but they aren’t required.

But I don’t have a problem with tiered service agreements, you see.

]]> By: RedBankTom http://bennett.com/blog/2006/06/blind-men-and-the-elephant/comment-page-1/#comment-296385 RedBankTom Fri, 09 Jun 2006 21:10:53 +0000 http://bennett.com/blog/?p=3918#comment-296385 Richard, I appreciate you taking the time to answer my questions. Would you be up for expanding on what you mean by this: “There’s no way the ISP should be deciding whether a given stream will have QoS or not’? Isn’t this what the discussion is about? ISP’s determining who gets to have those QoS type agreements? I don’t have a problem with your, or anyone’s, improved architecture of the internet; my concern is with entities such as ISP striking deals that allow QoS with their partners but not with competitors. Would an ISP be right or wrong if they choose to respect DiffServ markers of their partner's traffic but not of their competitors? Richard, I appreciate you taking the time to answer my questions. Would you be up for expanding on what you mean by this: “There’s no way the ISP should be deciding whether a given stream will have QoS or not’? Isn’t this what the discussion is about? ISP’s determining who gets to have those QoS type agreements?

I don’t have a problem with your, or anyone’s, improved architecture of the internet; my concern is with entities such as ISP striking deals that allow QoS with their partners but not with competitors. Would an ISP be right or wrong if they choose to respect DiffServ markers of their partner’s traffic but not of their competitors?

]]> By: Richard http://bennett.com/blog/2006/06/blind-men-and-the-elephant/comment-page-1/#comment-296384 Richard Fri, 09 Jun 2006 19:28:05 +0000 http://bennett.com/blog/?p=3918#comment-296384 Right, tshelton, a duopoly isn't the ideal configuration, but we already have multiple choices for TV with cable and satellite, and the potential of wireless broadband as well. We only have two choices for broadband in most markets because people are basically satisfied with them, not because it's technically impossible to add a third and a fourth. What do you suppose the practical limit is for broadband alternatives? I'd guess it's somewhere around three or four. Right, tshelton, a duopoly isn’t the ideal configuration, but we already have multiple choices for TV with cable and satellite, and the potential of wireless broadband as well.

We only have two choices for broadband in most markets because people are basically satisfied with them, not because it’s technically impossible to add a third and a fourth. What do you suppose the practical limit is for broadband alternatives? I’d guess it’s somewhere around three or four.

]]> By: Richard http://bennett.com/blog/2006/06/blind-men-and-the-elephant/comment-page-1/#comment-296383 Richard Fri, 09 Jun 2006 19:24:18 +0000 http://bennett.com/blog/?p=3918#comment-296383 Tom, in the QoS systems I work on, the client is the one who specifies the QoS parameters that the network enforces. This is the model we use with WiFi and with DiffServ on the Internet. So the whole picture is like this: you subscribe to your ISPs QoS service, and that entitles you a certain bandwidth allocation per day or per month. When you use a QoS-enabled service, such as TV streaming or VoIP, your software tells your ISP that you're going to use X bps in chunks of Y bytes. The ISP tells you OK, and you start your stream, and then you tell him when you're done. While your stream is blasting out or back or out and back, the ISP does an accounting and if you go over quota they either bump up your allocation or spread your stream across normal and high-priority queues. There's no way the ISP should be deciding whether a given stream will have QoS or not, because he can't even know what the application's requirements are unless somebody tells him. This model also should work if initiated by the other end, obviously. This doesn't strike me as scary, but I suspect ignorance of how QoS streams are setup and managed is one the main things that scares Neuts. Tom, in the QoS systems I work on, the client is the one who specifies the QoS parameters that the network enforces. This is the model we use with WiFi and with DiffServ on the Internet. So the whole picture is like this: you subscribe to your ISPs QoS service, and that entitles you a certain bandwidth allocation per day or per month.

When you use a QoS-enabled service, such as TV streaming or VoIP, your software tells your ISP that you’re going to use X bps in chunks of Y bytes. The ISP tells you OK, and you start your stream, and then you tell him when you’re done. While your stream is blasting out or back or out and back, the ISP does an accounting and if you go over quota they either bump up your allocation or spread your stream across normal and high-priority queues.

There’s no way the ISP should be deciding whether a given stream will have QoS or not, because he can’t even know what the application’s requirements are unless somebody tells him. This model also should work if initiated by the other end, obviously.

This doesn’t strike me as scary, but I suspect ignorance of how QoS streams are setup and managed is one the main things that scares Neuts.

]]> By: bonequark http://bennett.com/blog/2006/06/blind-men-and-the-elephant/comment-page-1/#comment-296308 bonequark Fri, 09 Jun 2006 14:10:36 +0000 http://bennett.com/blog/?p=3918#comment-296308 Richard, Another great article on the falacy bring purpetrated by Lessig and McChesney: http://econlog.econlib.org/archives/2006/06/net_neutrality.html Great writings, as always. Facts are such a funny thing. I wonder if Google is even smart enough to regret sending $ 4.76 million down the drain? Richard,

Another great article on the falacy bring purpetrated by Lessig and McChesney:

http://econlog.econlib.org/archives/2006/06/net_neutrality.html

Great writings, as always. Facts are such a funny thing. I wonder if Google is even smart enough to regret sending $ 4.76 million down the drain?

]]>