New Algorithm Boosts Network Efficiency

palegray.net writes "Researchers at the University of California have developed a new network routing algorithm that has the potential to significantly boost Internet traffic routing efficiency. This new approach focuses on the needs of dynamic networks, where connections are frequently transient. From the article: 'What the team did with their new routing algorithm, according to Savage's student Kirill Levchenko, was to reduce the "communication overhead" of route computation — by an order of magnitude.' For the technically inclined, the full research publication (PDF) is available."

Network efficiency algorithm

if($hostname==slashdot.org)
    connection.drop();

routers in perl?

That would make them blazing fast!

Re:Network efficiency algorithm

fixed it for you:

if($hostname=="slashdot.org")
        connection.drop();

What no compile/test cycle prior to submitting your post?

significant boost to algorithm

Don't use deep packet inspection for routing.

Interesting ...I'd think it would've been...

If( traffic == P2P || traffic == porn)
{
route_to_local_garbage()
}
else{
on_its_way()
}

Re:Interesting ...I'd think it would've been...

Actually, I bet spam outnumbers even the pr0n. Imagine a world without spam! All the pretty butterflies playing tag, and cute puppies rolling in the sunshine! Ahhh :)

fp

Amazing! I've never been able to get first post before, but with faster routing to slashdot.org, it was a sinch.

Re:fp

Maybe your ISP hasn't updated its routers yet...

Re:fp

Somebody give the guy at least a Score:1 Funny...

I mean, c'mon, "it was a sinch". Kind of like spelling better than most of /. is a cinch, only with more S's.

nearly as good?

so if my packets don't make it I know why. Not a skeptic but the Internet is already barely holding together and I'm not confident that "nearly as good" routing info can help. Of course if trying 2-3 times using this is still faster than first time hit using the old one then sure, why not?

Is this really new?

So, from reading the article, I see that the great leap forward here is "smaller routing domain in a link-state protocol leads to faster routing updates". But, looking at the existing link-state protocols, they were designed from the ground up with the ability to limit your routing domain manually so increase the convergence time and decrease memory footprint.

I guess that means the achievement here is to have a link-state protocol that automatically limits your routing domain by limiting propagation of routes. This however seems like it could lead to seriously suboptimal routing which is probably a bad idea in most network environments today.

Re:Is this really new?

The central point of the algorithm is to define bounds on when a routing change should be propagated. The point being that only an increase in routing efficiency above a certain threshold should be propagated. This disallows small fluctuations to have an impact on the wider network. He also shows that the impact of the propagation changes will be limited with respect to total network speed.

Re:Is this really new?

After reading a portion of the PDF supplied, I am actually far more satisfied that this is new research and not a restatement of fundamental network principles. The PDF has the equations where he proves a few simple criteria can define the scope of any network topology changes based on the difference in cost of the new route. This allows you to limit the recalculation of routes, blocking them from most of the routers where the recalculation would have provided no change in the actual routing topology.

The challenges he states are real challenges, and many modern networks are defined by the limits of the link-state protocols. In essence, this is like auto-summarization of prefixes in bgp, only applied to links in the link-state database - a possible slight loss in accuracy for a large boost in routing performance. This would allow the (faster converging) link-state protocols to scale to larger networks, rather then having to divide them into areas or use BGP to route between different sections of the network (resulting in loss of convergence time).

To the end user, this would mean that the internet would respond faster to outages, and better route around congestion on any one link.

Re:

By posting, you have removed your moderation, so you already did it yourself ;)

You can't post and moderate in the same story, otherwise you could post and then mod yourself, or mod down people who disagree with you etc.

Re:

Don't worry, I moderated him -1 troll to make up for it.

Re:Is this really new?

Ok, you seem to know what the hell is going on, so I'll ask you.

The summary talks about it being a huge boost to network efficiency and how it cuts overhead, but it seems like that wouldn't necessarily make a huge difference for most people's network use unless overhead is large and networks are hugely inefficient. I mean, if overhead is .0001%, then cutting it in half isn't going to give you too much of a boost in your ability to transmit data unless you're pushing around huge amounts of data and really need to squeeze every last bit of bandwidth. Right?

So I trust I'll get yelled at by someone if that logic is wrong, but just let me ask, what kind of benefit would this improvement actually yield? Do I get much better bandwidth, much lower latency, both? Or is it the sort of improvement than only a real network guru could love?

Patent?

So has the team applied for a patent? We wouldn't want just any ISP to be able to use this algorithm, would we? And if they don't patent it, one of the many patent-troll companies will, denying the researchers the right to use the results of their own work.

Re:

And unless they have millions to fight it out in court this is perfectly possible.

Re:Patent?

You must be new here.

Prior art is like kryptonite to the Patent Office.

Re:Patent?

You must be new here.
Redundant, I know, but when has the law ever stopped a patent troll?

Re:

This cannot happen. Once something is published, nobody can claim a patent on it at a later date... Even for the authors to apply for a patent, the application has to predate the publication.

Except in the US, where you have up to 1 year to file after initial public disclosure. Of course, the problem is that you can't get a foreign patent (because while the initial filing date is recognized by them, the fact it was published potentially nullifies any foreign patent. However, there are probably tons of except

The most important part

Finally, we argue that existing link-state protocols, such as OSPF, can incorporate XL routing in a backwards compatible and incrementally deployable fashion.

My first question upon reading the summary was, but is it backwards compatable... and they appear to answer that in the thesis statement. Looks like some good lunch reading here.

Re:The most important part

They meet the âoecentral challengeâ of determining which updates are important and which can be suppressed by using three rules for update propagation, said team member Ramamohan Paturi.

1. The routing algorithm may not injure the network or, through inaction, allow the network to come to harm.
2. The routing algorithm must obey orders given to it by human beings, except where such orders would conflict with the First Rule.
3. The routing algorithm must protect its own existence as long as such protection does not conflict with the First or Second Rules.

Seems pretty foolproof to me.

Re:

Seems pretty foolproof to me.

Nah, you just present it with a situation where acting will harm one human and failing to act will harm another. Then it jams up and starts vibrating and sparks shoot out of its ears. (Or at least that's how it works for robots. To be honest I don't know where a routing algorithm's ears are, but this seems as good a way as any to find out.)

A new Al Gore rhythm

A new and improved Al Gore rhythm would dramatically boost network efficiency. Since he invented the internets, and actually routes every single packet on the internets by hand, if he learned how to work in a syncopated rhythm the efficiency of the network would nearly double.

Check and mate!