Comments on: XP: Prevent QOS (Quality of Service) from Reserving Bandwidth http://www.tech-recipes.com/rx/254/xp-prevent-qos-quality-of-service-from-reserving-bandwidth/ Computer and technology tutorials and guides Sun, 22 Nov 2009 02:44:39 -0800 http://wordpress.org/?v=2.8.6 hourly 1 By: Anonymous http://www.tech-recipes.com/rx/254/xp-prevent-qos-quality-of-service-from-reserving-bandwidth/comment-page-1/#comment-1846 Anonymous Wed, 11 May 2005 07:54:31 +0000 guid-fix-me!#comment-1846 Well, in engineering nothing is really true of false but built more around "use cases". From what I have read, this feature mearly balances the amount of bandwidth between application/connections on the local host. For example: you start to download 1 mp3 that is 3 meg. Then you start to another 1 meg mp3. In the old style WIN TCP stack (network connection) the majority of the band width is use by the first download while the second download uses only a little bandwidth needs to wait for the first to finish before it can increase it's connection capacity. In XP, the first connection will use 100% until the second is started. At that time the second download request starts, 20% or so of total bandwith is instantly allocated to the second program and after that a round robin (1-2-3-1-2-3-1-2-3-1-2-3) starts to occur giving slightly more band width to the application until things are equal. Microsofts technical white paper is just that, a technical document that not only mixes network engineering but all also client TCP stack design details. I think the people that made the recommendation to turn off QoS should probably brush up on their TCP/IP basics. The paper is very clear if you understand how TCP retransmission occurs it makes perfect sense. TCP is connection based protocol. In other words there is some checking that you get all the requested data. As the data streams to you, some times bits are lost along the way. This is generally called packet loss or just loss. Each packet has a number (1,2,3,4,5,6,7,8,9) so if a packet is missing the recieving side will ask for that packet number again 1,2,4,5,6,7,8.... Oh, can I have #3 again plase). We call this "retrans" (retransmission). In the first "old style" example the the first mp3 soaks up the bandwidth but the second does get a chance to make the request but lwill ose a few of the packets due to saturation caused by the first download. Retransmission occurs which means that even MORE traffic on the network which isn't a good thing. There are a lot more reasons why QoS in XP (and 2000) is a good thing. Disregard the artical and keep it turned on. Well, in engineering nothing is really true of false but built more around “use cases”. From what I have read, this feature mearly balances the amount of bandwidth between application/connections on the local host. For example: you start to download 1 mp3 that is 3 meg. Then you start to another 1 meg mp3. In the old style WIN TCP stack (network connection) the majority of the band width is use by the first download while the second download uses only a little bandwidth needs to wait for the first to finish before it can increase it’s connection capacity.

In XP, the first connection will use 100% until the second is started. At that time the second download request starts, 20% or so of total bandwith is instantly allocated to the second program and after that a round robin (1-2-3-1-2-3-1-2-3-1-2-3) starts to occur giving slightly more band width to the application until things are equal.

Microsofts technical white paper is just that, a technical document that not only mixes network engineering but all also client TCP stack design details. I think the people that made the recommendation to turn off QoS should probably brush up on their TCP/IP basics. The paper is very clear if you understand how TCP retransmission occurs it makes perfect sense.

TCP is connection based protocol. In other words there is some checking that you get all the requested data. As the data streams to you, some times bits are lost along the way. This is generally called packet loss or just loss. Each packet has a number (1,2,3,4,5,6,7,8,9) so if a packet is missing the recieving side will ask for that packet number again 1,2,4,5,6,7,8…. Oh, can I have #3 again plase). We call this “retrans” (retransmission). In the first “old style” example the the first mp3 soaks up the bandwidth but the second does get a chance to make the request but lwill ose a few of the packets due to saturation caused by the first download. Retransmission occurs which means that even MORE traffic on the network which isn’t a good thing.

There are a lot more reasons why QoS in XP (and 2000) is a good thing. Disregard the artical and keep it turned on.

]]> By: Anonymous http://www.tech-recipes.com/rx/254/xp-prevent-qos-quality-of-service-from-reserving-bandwidth/comment-page-1/#comment-1083 Anonymous Fri, 26 Nov 2004 22:30:44 +0000 guid-fix-me!#comment-1083 By Microsoft "There have been claims in various published technical articles and newsgroup postings that Windows XP always reserves 20 percent of the available bandwidth for QoS. These claims are incorrect." also from Microsoft "By default, programs can reserve up to an aggregate bandwidth of 20 percent of the underlying link speed on each interface on an end computer. If the program that reserved the bandwidth is not sending sufficient data to use it, the unused part of the reserved bandwidth is available for other data flows on the same host. " This tech-recipes article is not true on QoS. By Microsoft

“There have been claims in various published technical articles and newsgroup postings that Windows XP always reserves 20 percent of the available bandwidth for QoS. These claims are incorrect.”

also from Microsoft

“By default, programs can reserve up to an aggregate bandwidth of 20 percent of the underlying link speed on each interface on an end computer. If the program that reserved the bandwidth is not sending sufficient data to use it, the unused part of the reserved bandwidth is available for other data flows on the same host.
”
This tech-recipes article is not true on QoS.

]]> By: Anonymous http://www.tech-recipes.com/rx/254/xp-prevent-qos-quality-of-service-from-reserving-bandwidth/comment-page-1/#comment-106 Anonymous Wed, 25 Feb 2004 11:17:39 +0000 guid-fix-me!#comment-106 If you read the recipe above, it appears that the original author agrees. If you read the recipe above, it appears that the original author agrees.

]]> By: bob http://www.tech-recipes.com/rx/254/xp-prevent-qos-quality-of-service-from-reserving-bandwidth/comment-page-1/#comment-104 bob Wed, 25 Feb 2004 09:34:51 +0000 guid-fix-me!#comment-104 This is all a myth. Unless you are on an Active Directory domain with QoS enabled applications then it does nothing! This is all a myth. Unless you are on an Active Directory domain with QoS enabled applications then it does nothing!

]]>