Who is in Charge of IPv6 Packet Priority?

OwenF asks: "Here's a question. IPv6 is well on its way to becoming the next Internet packet routing standard. One of the big "features" many proponents of IPv6 enjoy pointing out is that IPv6 packets will include "priority" and "discardabillity" info in their headers. Right now, using IPv4, packets are shuttled across the net on a first-come first-serve basis, and when network congestion gets too high, servers start dropping packets, also on a first-come, first-serve basis. With IPv6, however, packets can be encoded "Top Priority" and/or "Do not discard" so that they recieve special treatment at the hands of the routers. What I want to know is, who's going to have control over these header flags? Can I make all Web server's transmitted packets "Top Priority/Do Not Discard," or is this just another way for big-name commercial web sites (e.g. those using streaming audio) to take more control of the net and force other user's packets into digital heaven? "

Who set priorities? You do

[Anonymous Coward]

...or at least you can, for anything originating from or passing through a box you own. I don't imagine that anybody else will feel much obligation to respect your priority settings. I can imagine ISP's charging extra for not resetting DnD bits, then failing to pay their upstream's equivalent fee.

ISP: "We're multihomed with redundant T3's!"
Me: "Cool! What's your setup? BGP?"
ISP: "uhhh... What's BGP?"

Let the market decide

[VP]

Remember - you heard it here first! :-)

One idea that I've heard (and I don't know whether to like it or fear it), is that packets will negotiate their pritority by "paying" a toll to the routers. I am not aware of anything out there implementing this, but it is at least a very interesting problem to study, and model.

Re:It is kind of interesting.

[lotaris]

There is, IPv6 will have to be phased in:

Take a look at http://www.ipv6forum.com for more info.

It is kind of interesting.

[noweb4u]

Ok, so what happens to boxen I don't upgrade? I know that there is some backwards compatibility, but what if the nameservers go IPV6? How am I going to resolve names?
Perhaps there is a forward compatibility solution I don't know about...
-Paul

Finances

[jfernie]

QoS schemes always include paying for higher quality service. In general, QoS schemes assume that Network-To-Network Interfaces (NNI's) will respect the priority of packets, and simply expect financial compensation for that privilege. You're upstream provider will have to pay more for routing higher quality traffic out to the internet, and he'll pass that bill on to you. So in the end, you are the one regulating priority.

I haven't looked at the IPv6 specs [isi.edu] in a while, and when I did last I wasn't really paying attention to that specific point, but every method I studied back in my networks class assumed priority was respected in exchange for cash.

ipv6 ipv4

[hodeleri]

You'll have to RTFM for details (RFC-Editor [rfc-editor.org] or your local ipv6 man page) but there is an entire suite of methods for providing the necessary junction between the two. One of the nice things is you can implement ipv6 on your routers and big boxes and your dumb-as-a-rock MS boxes that run ipv4 won't know the difference. (knock on wood)

--
Eric is chisled like a Greek Godess

IPv6 / QoS / Policy based routing

[velkro]

From what I know, it will work similar to this:

ISPs will have control over QoS within thier own networks. So when a packet from ISP 1 goes thru ISP 2's network, ISP 2's routers strip the QoS tagging done by ISP 1, and put ISP 2's QoS information in the header.

If I remember right, this can be done on both a source & destination basis, so transit ISP's will probably set the priority high on traffic originating / terminating within thier networks. As least I know I would.

Now, this doesn't stop ISP 1 from paying ISP 2 extra cash to give them a higher priority - but then again - we need something left open to conspiracy theorys...

IPv6 and Quality of Service

[PatJensen]

The new functionality that this provides will alllow for Quality of Service changes across your network, and your providers level at a more granular level.

Today, with IPv4, we have different types of queueing in place to ensure our data arrives safely. However, using router features like custom queueing can take a large amount of processing power to make sure that certain traffic is permitted, and certain traffic is dropped.

Using the new IPv6 features, the router can automatically flag entering and exiting packets inside of your network with whatever priority your routing policy allows. For instance, Voice-over-IP packets can be routed first, giving less latency and better quality in your calls.

This can bring about various changes at the NSPs, but is least likely to be bad for the consumer and as IPv6 starts making inroads .. I believe this functionality will ensure your packets will arrive quick and intact.