NGN wiki page
Finally posted this. Would appreciate review of the maturity/impact assessment numbers, also I suspect we can do better on quotes and experts.
Meantime, Tom and I began a discussion by email we thought would be better opened up to those assembled here: what will the impact of NGNs be on the Internet?
Tom wrote that he thinks NGNs will either kill or cure the Internet: either it will force the network to grow up and become more mature and reliable (more disciplined) or it will I guess slice stuff up and destroy it. (To me, the first of those options is also a kind of destruction, as having the phone companies control the Internet would turn it into a controlled network like the old telephone networks and unlike the Internet we’ve known so far).
If Tom gives permission, I’ll copy and paste his email in here; there were more thoughts.
I think I’ve been living in the hope that the telephone companies wouldn’t gain total control; I do wonder, though, about the reliability of these NGNs. I remember Matt Blaze saying at a CFP years ago that no computer anywhere was ever engineered to the standard of the old telephone networks. Will we be able to make up for that with massive redundancy?
wg
August 12th, 2007 at 8:09 am
Hi Wendy,
Thanks–and yes, please feel free to post our correspondence. Your wiki entry is really quite good–I hope visitors will have a look and come back with reactions. I think it’s important to separate impacts from discussions of network neutrality/ownership here, although they do bleed into each other from each direction.
I’m particularly interested in a discussion of the likelihood of P2P or CDNs or massive online games impacting overall performance, and whether differing types of NGNs have differing levels of impact. Other commenters? Wendy?
August 13th, 2007 at 3:22 pm
OK. Here’s what Tom said by email:
>>First, ngn’s will either kill or cure the Internet. Either their
requirements for performance and low latency will bring needed levels of
discipline to the dumb pipe chaos currently extant, or their slicing and
dicing of the Internet will cause rack and then ruin.
P2P and other overlay networks are the real issue, as they degrade
performance at the margin of IP, and can constitute a threat to overall
packet performance.
Tiered access with pay for performance is the logical goal–the key will
be guaranteeing SLAs for free users and determining who pays for the
upper level networks. Those wanting to pipe video down from a CDN have
the biggest axe to grind with P2P networks, and they should be allowed
to fight it out, peferably in a World Wrestling Leage format.>>
And
>>The problem with overlay networks is that they corrode traffic
performance on the margins by using differing protocols on top of
TCP/IP. I’ll send you something from my home computer in a minute.
If you look at what BT is doing (building one of the world’s largest
CDNs), you get an idea of the scale of mass downloads of hi def
video–remember that YouTube’s 105 million daily downloads of low def 10
minute videos caused capacity issues at first. >>
wg
August 13th, 2007 at 3:32 pm
It’s worth pointing out that P2P and video aren’t the first applications to swamp available bandwidth - the Web itself nearly killed the Net when it first took off. I think it’s only timing and slower growth that stopped VOIP from doing it. I also think the threat, such as it is, from P2P and other bandwidth-sapping services yet to be invented can be handled by for example priority handling - BT’s 21CN is using MPLS to ensure QoS in voice calls (and later, presumably paid video), but P2P doesn’t need to have immediate delivery guaranteed - does it matter how long it takes to download a file? If you want guaranteed service you pay more.
Of course, the problem with all this is that it builds into the network something we haven’t had before and something that net neutralists are fighting over - classes of traffic. I think that may be acceptable as long as it’s not content or ownership that determines priority.
wg