« Ben West's coverage of ISCWN | Main | Universal white spaces - moving beyond the TV bands »

August 24, 2010



The Internet became what it is today, because of the IP protocol. It's just a way to connect computers with each other. What's the problem with that? I don't care which protocol is used to send and receive my data, as long as I can send and receive anything to/from anyone.

You are able to connect to others using the IP protocol because your Internet provider supports that protocol. If you want to run your own (lower) protocols, your ISP should support that protocol too, because else it wouldn't understand what you are trying to do.

There's no difference between being forced to use PSTN and run other protocols over that or being forced to use IP. They're both protocols and everyone has to use them to work.

What you want is just a big chaos of thousands of small networks each running their own protocols and unable to communicate with each other. Not a good plan.


I would like to get a clarification on common carriage that you advocate. You suggest that we should get equal access to the dark fiber. Since ultimately we would like to have networked coverage, I am presuming that the dark fiber will be terminated at some "network service provider". Am I right?

If we get access to the dark fiber, who will maintain the fiber? The fiber provider will not have access to management information to provide maintenance. The NSP will not have access to the physical cable. Won't we end up being kicked around two who will be busy passing the ball around?

You mention T1 as an example of access to bit-stream. But there we had different kinds of access: channelized, D4 and ESF; but all with NA format. At first blush dark fiber looks like a bit untenable. But you also suggest that it is being done in Europe. How do they address these issues there?


Can I paraphrase? "I used to be able to layer a protocol on top of voice circuits. However, layering a protocol on top of an IP data stream is entirely different. I want dark fiber."

T1's and modems were not the equivalent of dark fiber. They were the equivalent of the raw IP connection. Everything you sent down the T1 was layered on top of the standard voice handling logic in the network.

I fail to see the difference between layering a new protocol on top of a T1 circuit and layering one on top of an IP connection.


Hi WeAreGeek, Thanks for commenting.

My point was (and is): if you can only innovate above the IP layer (because access to the lower layer functions is effectively denied), then innovation at those lower layers (which possess their own protocols) is likewise denied.

You may only want to use applications above the IP layer, but there is plenty of room for innovation at lower layers. We some of this in corporate LANs where access to lower layers is available. And some of this lower layer innovation might even benefit you ten years from now. After all, the Internet could never have been deployed if we hadn't had access to lower level transport in the 1980s and 1990s.


Hi Aswath,

I do advocate open access at lower practicable layer which in many cases is dark fiber. If we grant monopoly access to our rights-of-way we expect something in return. I am suggesting that open access to dark fiber would be better in the long term. It's worth studying Stockab A.B. in Stockholm. They are a dark fiber provider (since 1994) and, while it took 6-8 years to evolve, their existence has resulted in hundreds of speciality ISPs and private corporations leasing dark fiber in the greater Stockholm area. It's also meant that 100/100 Mbps Internet connectivity in Stockholm costs 95 Kroners per month (~$12) if it's not bundled in your apartment rent.

There are also many examples of dark fiber being available in the Internet backbone. It's only in US access links that we have a (now effectively unregulated) monopoly and have to fight over whether there is open access at layer 2 (historical) or layer 3 (net neutrality).


Hi Jason,

I never said T1s and voice circuits were the equivalent of fiber. I said we historically had open access (common carriage and regulated rate of return) at layer 2 and now we were fighting to get open access at layer 3 (net neutrality). At a minimum, we should fight for what we had.

Separately, I suggested that, if you want to fight for something other than what we had, the fight should focus on lower layers (e.g. dark fiber), not higher layers (Net Neutrality). But these were two different points that I included in one posting.

As to tunneling everything over IP in the access network, that doesn't cover many useful things you'd like to do. My VoIP example is just one case that is real today. IP access services from the duopoly do not include differential priority. As a result, enterprises must buy two separate access links in order to give priority to voice. This is trivial to handle at layer 2 or at layer 3, but it's not part of the IP access service that is offered by the duopoly and, without lower layer access, it never will be.



With a T1, you didn't have differential priorities either. It was all delivered with one priority. There is nothing preventing anyone from running their own network on top of the IP "circuits" that they run between offices. In fact, many companies already do this. Actually, many carriers are doing this, layering Internet traffic on top of an internal IP network.

I still fail to see the difference between an IP link between two sites and a T1 connection. Both have delivery guarantees, and both provide a bitstream service (both require routers) which the customer is able to use to layer their own network on top. QoS is up to the sender to enforce, it is up to the carrier to meet their SLAs.

Arguing for common carrier is good. Arguing for dark fiber to implement your own network by comparison to a T1 or modem? Not so good, false analogy. You should have compared dark fiber to the unconnected POTS lines that you could purchase for alarm connections. They were very rare beasts back then too.



With T1 you can introduce priorities and many people do. The point is, with T1, you are operating below layer 3 (below IP) and you can put your own equipment at both ends of the link. T1 trunks allow CLECs to prioritize traffic so a single T1 trunk provides both generic Internet access and preferred bandwidth for Voice traffic. Once upon a time (prior to the late 90s) T1s were both common carriage and regulated rate-of-return. Today T1s are an effectively unregulated monopoly so T1 pricing is still based on the cost of voice circuits in the early 90s (ignoring 15-20 years of Moore's law). forgive me for emphasizing common carriage when we need both common carriage and regulated rate of return (at the lowest feasible layer).

The point is, with T1, you are operating below layer 3 (below IP) and you can put your own equipment at both ends of the link. Your comment makes me realize that I also didn't emphasis the other thing about lower layer connections like T1, which is the ability for competitive carriers to get at both ends of the circuit means they can deply new gear and thus build innovative new networks.

Today we have lower layer access inside the enterprise (where you can run your own cables or fibers or anything else, and we have lower layer access in the Internet backbone (you can lease dark fiber, wavelengths or Ethernet trunks between Boston & NYC or Los Angeles). But we don't have lower layer access (at any reason cost) in the access link. Interestingly enough we do have innovation at lower layers, both within the enterprise (dozens or hundreds of applications) and within the backbone (just think of Akamai, Google caching, wavelength routing and dozens of other innovations).

Network neutrality just means we'll never have innovation below the IP layer in the access network.

I sorry if my discussion of dark fiber muddied the waters. My primary point was and is: be careful at which layer you seek open access because you will preclude innovation below that layer. Network neutrality is open access to layer 3. Common carriage with regulated rate of return gave us open access at layer 2 (or layer 1 depending on how you argue it). At a minimum, I'd prefer a return to common carriage with regulated rate of return at layer 2 (or layer 1), i.e. to what was supported by laws and regulations throughout most of the 20th century.

As a separate comment, if we wanted to fight for something different that what we had in the 20th century, we should be fighting to enable more innovation, not less. I.e., we should be fighting for open acces at lower layers (like dark fiber & bare copper) rather than fighting for network neutrality (layer 3 and above).


Saying that there won't be innovations below the IP layer is demonstrably false, because we are continually seeing improvements below that layer and we don't have network neutrality now. Carrier core networks are changing on a 5-10 year basis. Mobile networks are being replaced every 5-7 years, with substantial upgrades every 2-3 years. The upgrades that are happening right now in the mobile network are all below the IP layer - all to improve IP performance.

We're seeing DWDM, 100G ethernet, MPLS, and ethernet trunking all looking for traction in the carrier space. Again, all below the IP layer. I just recently saw a slide from a telco service arm which had:

Telco ethernet point-to-point links
MPLS routing
Telco IP network
Service Provider IP network (encrypted VPN)
Customer external IP network
Customer internal encrypted VPN network (also IP)

All on one link. So, in that situation, what does it mean to be able to get below IP? IP is running on top of IP 4 layers deep!

Even better, they could give you a T1 line and you still wouldn't be below IP:

Telco ethernet point-to-point links
MPLS routing
Telco IP network
Service Provider IP network (encrypted VPN)
T1 link between customer locations
Customer internal encrypted VPN network (also IP)

So, they've given you a T1 line, which given the definitions you've given would allow you to innovate below IP, and you're still running on an IP connection. Therefore, there is no difference between the IP connection and the T1 line.

It's all bits, and unless you really need access to the physical layer (say to implement quantum encryption), you can innovate at whatever layer you want. It all comes down to the SLA agreements that are negotiated with the carrier and service provider.


If you recall, there was a time back in the 80s that ISDN was being touted by international standards organizations (e.g. OSI), as *the* way to go about standardizing international data communications. The Internet Protocol (IP), developed under DARPA initiatives, was an upstart concept practiced by a relatively small academic and research community.

Why did the Internet Protocol thrive and prosper, and ISDN fall by the wayside, only to fade into obsolescent obscurity ? One of the reasons, of which there are many, is that ISDN is/was a connection-oriented protocol, and IP is a *connectionless* protocol. ISDN requires that a virtual circuit be set up and maintained for every on-demand new connection ('call') between any two points on an ISDN switching/routing network. If, for any reason, that fixed virtual circuit encounters a transient network error, the connection is, yep, BROKEN, and you must reconnect and try to recover whatever the state of the session might have been.

On the other hand, IP is connectionless, as information is routed in packets with addresses and not over virtual circuits which have been instantiated in all of the routers along the way in a given virtual circuit. An IP packet can be delivered any number of ways through the network, and -- best of all -- packets can simply be discarded if and when routers or gateways become congested.

Finally, keeping an ISDN virtual circuit open from end-to-end required maintaining a constant allocation of resources in the network, much like a traditional switched voice network call. If the session using the connection is idle, you are still running up the taxi meter. With IP, no activity means no additional network resource cost.

Why do I mention this historical example of the rise of IP over ISDN ? Your desire to get at the bitstream is analogous to the patchwork of point-to-point connections which are required in virtual circuit switching. Getting access to a lower-level data link is no guarantee of the success of a network initiative, particularly one that ( I assume) you think should be a publicly available resource for general purpose use.

You would have to develop your own network routing protocols for all of the traffic on a ' bitstream innovation ' that you wish to have the freedom to develop. The problem is not the economics of the pricing of getting access to dedicated bandwidth of the multiplexed bitstream -- the problem is all of the protocol work which you would have to do to make such access an information routing facility which would support any point in the world connecting reliably with any other point in the world -- which IP already does.

Using lower level data links for dedicated applications is still possible. However, your contention that access to this layer is needed to further innovation in a communication network serving the planet is completely misguided, misinformed, and doomed to failure before you begin. You will be left alongside ISDN in the dustbin of history.

Sorry, no go, complaint without merit, case dismissed :)


P.S. If the reason you seek to preserve access to lower level data link bandwidth is to use such access to offer competitive IP networks bridged by these links (e.g., in the CLEC context to connect the CLECs), then my criticism of appearing to seek network protocol alternatives to IP would need to be tempered.

This is not all that apparent from your discussion, however -- and I am only surmising that this could be a possible rationale in view of your efforts in alternative community networks -- but I at least thought I should mention that if that is in fact what you were driving at.

Kenneth R. Carter

I would like to offer a tiny clarification. The common carriage obligation is independent of monopoly. The obligation attaches when a firm holds itself out in common carrier businesses, regardless of market concentration.


Of course there will still be innovation below layer 3, but it won't benefit US fixed line customers. There will still be innovation below layer 3 in those places where their is still access below layer 3, for example, in enterprise equipment, in the Internet backbone and in access circuits in other countries. But US consumers won't have access to any innovation below layer 3 in the fixed access network or in most of the "middle mile" network.


Thanks for the qualification TahoeBlue as I am no lover of ISDN (or ATM). I am precisely interested in the kind of innovation that originally produced the Internet and the kind of (enterprise-based) lower layer innovation that is driving down the cost and increasing the flexibility of optical gear. I have been an advocate for dark fiber and condominium fiber (and open spectrum) and indeed I am currently running a start up ISP, netBlazr (so far still local to the Boston area).

Prior to the mid-90s, we had open access at layer 2 (and sometimes layer 1). We would benefit from open access to dark fiber and raw copper pairs, i.e. layer 1 or 0. Why ask for open access at layer 3 (something new that's less powerful than what we had)? If you want an easier fight, fight for what we had. If you want to fight for something new, fight for access to dark fiber!

Jordan 13

The administration of the law can never go lax where every individual sees to it that it grows not lax in his own case, or in cases which fall under his eyes. (Mark Twain, American writer)

dooney and bourke outlet

Thank you for sharing, you're just for me is very useful. I really like your post.

Sam Smith

One of the best designed blogs. worth visiting. great tips and tutorials. Keep it up !

The comments to this entry are closed.

My Photo

Search this Blog

Subscribe by Email

March 2014

Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          


Site Meter

Upcoming Travel & Conferences

Twitter Feed

Become a Fan