Quick Links

 

• Frequently Asked Questions - What is a NAP?

• Frequently Asked Questions - Technical Information

 

Q. What is a NAP?

(This artical is from our sister company KC NAP)

A. When you connect to the Internet, what you are really doing is connecting to your Internet Service Provider (ISP) and their network, a very tiny subset of the Internet. What you want is all the networks, not just Sprint's, ATT's or MCI's piece of it. Because of this all ISP's go to the public network access points in Chicago, Dallas, Washington DC (Virginia) and San Francisco. They also peer at many private points (P-NAP's) as the public points often get overloaded. Close to 40 major P-NAP's worldwide now are in operation.

Kansas City now has one of these private NAP's. Our local traffic 3 years ago was about equal to 6 DS/3's now it is more than 250 times this volume, with many private companies now having their own DS/3's. As the distinction between voice and data networks has fallen, and the rise of technologies like Video Conferencing and Voice over IP have grown the need has become more pressing.

However since no one entity is responsible for these types of interconnections, they only happen when people who organize themselves behind a good idea press it forward. To this end CTC and several of our key customers and vendors have extending themselves to organize for this purpose forming the KC NAP, LLC. We welcome you to join us. Since this is a complex subject, we have provided basic information in a question and answer format.

 

Q. Why Does Kansas City needs a Private NAP?

A. The Internet is becoming an indispensable tool for several thousand companies and hundreds of thousands of individuals in the region. These customers are serviced by dozens of ISP's and hundreds of corporate networks in the metro area. All of these ISP's are currently backhauling their traffic to The Ameritech NAP in Chicago (the largest) or the SBC public NAP in Dallas, or a private NAP in Denver. A very few are passing their data to Public NAP's in Washington DC, Miami or San Francisco. While some private peering agreements exist, almost all located elsewhere, the vast majority of cross-town traffic ends up going thousands of miles out of its way.

 

Q. Why is this bad, doesn't the Internet traffic move at the speed of light?

A. Almost, but the real issue is the latency (time delay) caused by even these very fast transits. A few thousand miles of travel can add 20 to 40 ms of latency to your Internet times. Add to this 10 to 35 ms of router, switch and device overhead associated with the average 17 hops it takes to get from one ISP to another and the 25 to 100 ms of delay that can be caused by overloaded internet pipes and you end up with times between 55 ms (good) and 175 ms (bad). Naturally the fewer hops (routers) you go through the better. Travel by way of the Public NAP's has typically between 10 and 25 hops involved, with averages in the 17-hop range. Travel via a Private NAP typically cuts these numbers in half.

 

Q. So your saying that not only is speed increased due to the Private NAP not sending signals thousands of miles, but that less infrastructure must be passed through as well?

A. Absolutely. While none of us begrudges the long haul fiber networks there piece of the pie, it shouldn't be necessary for our local Kansas City traffic from ISP A, to go to Chicago, courtesy of Sprint, then go to Dallas by way of WorldCom, then come back to town, courtesy of Quest, finally to be delivered to ISP B. Not only is this wasteful of capacity, it also involves far more points of potential failure. With a private NAP in place ISP A could exchange data directly with ISP B. And in some cases data could move less than a few hundred feet in the same building or a few thousand feet, in a few block area rather than a few thousand miles!

 

Q. What do you mean by this?

A. Kansas City has the vast majority of its ISP's and corporate networks physically located in the downtown loop. The vast majority of fiber, high speed copper and fixed wireless in the area actually lands within about a 6-block radius of 11th and Grand. In fact 3 buildings house over 60% of area ISP's or a large portion of their infrastructure. These are the Bryant Building at 1102 Grand, Switch and Data at 1025 Grand and the Oak Tower at 324 E 11th. When you consider that SBC and ATT are also in this same area and they act as the interconnects for most of the area regional customers by copper, it is easy to see that potentially one half of local area Internet traffic could be kept local.

 

Q. What is the major benefit of doing this?

A. Much better performance, and much lower cost. All ISP's base their elastic pricing on two principle components: the cost of transit; and the cost of transport. Transit is the handoff from one ISP or network to another; Transport is the cost to get from point A to point B. Sometimes this is listed as the cost of the port (transit) and the cost of the local loop or pipe (transport). Because of advances in Gigabit Ethernet technology, fiber, wireless and free space optical, not to mention the very short distances involved, the cost of transport can be almost eliminated, or drastically reduced. Savings up to an order of magnitude are possible. Even transit is greatly reduced, as fewer handoffs are likely, and the ones that are made cost correspondingly less to make due to short distances and the larger interconnects that allow economies of scale.

 

Q. What do you mean by economies of scale?

A. Strange as it may seem the cost to provision Fast Ethernet, (100Mbs) and Gigabit Ethernet, (1000 Mbs) are much lower than the costs to connect DS/3 (45Mbs) or OC-3 155Mbs or OC-12 (622 Mbs). However Ethernet connections can only be used very short distances (in the order of a few hundred feet). Ethernet can currently be scaled up to 10,000Mbs that's roughly 222 DS/3 connections. Ethernet is very easy to provision in fast switching and routing environments. So if you can bring all the parties together in close proximity, as we can do here in Kansas City, it would be possible to use these much faster Ethernet connections. (Optical conversion will operate quickly enough to allow all the downtown loop areas mentioned to interconnect inexpensively).

 

Q. What new services will this allow?

A. The most compelling of these is Voice over IP, and higher security services based on VPN tunneling. Additional services such as real time monitoring, security video, video conferencing and real time process control all become remarkably efficient if latency can be held under 30-50 ms or so. When you think of the importance to this area's economy of emerging technologies such as life sciences it becomes crystal clear how real time experimental modeling and process control could be used by the scientific community to effectively handle and transmit mission critical data over networks that are more reliable and more secure.

 

Q. What about the costs of building a private NAP?

A. Consider the costs of shipping our data several thousand miles needlessly? Private NAP's historically have lowered costs, improved efficiency, and made regional networks more valuable. Networks become more valuable when the number of endpoints increases, and the speed and reliability rise. Private NAP's or P-NAP's do this very thing. Consider the promise of Voice over IP, using the same converged networks for voice and data. Traditionally the Public Internet has been considered to slow and unreliable to make Voice over IP work well. For the layperson VoIP starts producing unsatisfactory results when latency rises much over 120 ms. (about an eight of a second). Kansas City could be the hub of a working VoIP nationwide network, if data exchange happened here in the middle of the country rather than being sent from one coast to another just to cross network boundaries. Kansas City is roughly 30ms from any other point in the USA and Canada, and roughly 50 ms to anywhere in this hemisphere. Given the physics of the Internet a KC P-NAP could be instrumental in making an overlay network optimized for high priority VoIP traffic highly successful here.

 

Q. What is an overlay network?

A. A network built in parallel to or over the Internet networks currently in place. Overlay networks are actually quite common and are normally built for speed, other performance or security issues. One of the greatest strengths to a vendor neutral P-NAP is the ability to allow corporate networks, banking networks, financial networks and government networks to transfer data regionally at low cost and not be exposed to the general Internet with its inherent weakness in security, and performance, but still able to closely match or even exceed the Internets low cost of operation.

 

Q. Give an example?

A. The New HIPPA (HEALTH INSURANCE PORTABILITY AND ACCOUNTABILITY ACT OF 1996) regulations make it difficult for medical records to be exposed to the general Internet at the same time medical and insurance companies are looking to the Internet for the kinds of savings that are so badly needed right now. Both, to contain costs and increase service. With only minor upgrades to security and procedures it would be possible to imbue a new regional P-NAP overlay network with the security safeguards that would allow all the benefits, without the risks or costs needed to make the Internet HIPPA safe. Consider what this would mean to the extensive Hospital, Collegiate and Life Sciences community now having to live with these new regulations.

 

Q. If the benefits of a P-NAP are such a good idea why hasn't anyone done it yet?

A. The answer is they have many times over. Much smaller regions and Cities have already built P-NAP's. Including cities like Austin-NAP, and Louisville-NAP. Even Kansas Cities own Sprint runs one of the largest P-NAP's in the world, the one in New York City. The Public NAP's started out as private ones, the Chicago NAP, the largest NAP in existence where over 108 providers exchange data at OC-12 speeds was largely built as an Educational NAP. (The Chicago NAP has provided the hub for the big ten universities and is sometimes called the Midwest Research and Education Network (MREN). This consortium was formed in 1993 and has been a key partner in Ameritech's data services.)

 

Q. So why not in Kansas City?

A. Why not indeed! Not only is it time, but past time to build one. It was not done before this due to the combined factors of the Tele-Comm industry meltdown and the lack of venture capital in the region. The larger players (since they all have large fiber optic backhaul networks) have not championed the cause as it would undermine their revenue base in the long haul market, or the not-invented-here mantras have slowed the development of these ideas.

 

Q. Why haven't the big telephone companies done it?

A. All the big Public NAP's currently in existence are run by the long distance, fiber, or telephone companies, in fact the former Ameritech's site in Chicago and the one in Dallas are owned or run by SBC. Most if not all of these networks were designed to originally carry voice, not data, and one of the first technologies to take off from a KC P-NAP would be Voice over IP (VoIP) a major threat to the traditional Telco economic and power base. Considering that the other NAP's in the country are run by PacBell, Sprint, Bell South and WorldCom there has not been much economic reason to have the ILEC's do it.

 

Q. Don't the Telco's provide reasonably priced data circuits?

A. Not necessarily. In fact until very recently a DS/3 to the Internet would cost a company as much as $20,000 a month. Just the local loop for a DS/3 is several thousand dollars. But as even very small ISP's have proven, this cost structure doesn't need to be in place. For example CTC sells wireless DS/3 data service (an extension of Ethernet) in the downtown area for $2500 to $4500 dependent on term, less than 1/4th the cost of the traditional Telco's offering. When you consider that some T-1's (1/28th) the speed, cost $1,200 a month it is now possible for downtown businesses to buy really big pipes at a fraction of their former costs and often not much more than they were paying for 1.5Mbs of service. When 45 Mbs pipes come down in price to the level of the lowly T-1, vast new arrays of service become possible. In fact Ethernet delivery has proven to be so dramatic in lowering costs companies like Cogent and Yipes are now delivering gigabit Ethernet speeds for the price tag of the old DS/3.

 

Q. If the costs associated with Ethernet delivery are so much lower why hasn't everyone done it that way?

A. Ethernet delivery is reasonably new at these speeds; most of the older networks were built based on ATM and frame relay. Since Ethernet has enjoyed the economies of scale that the corporate LAN (Local Area network) has enabled, it is now possible to build WAN and Metro WAN structures using the same technology and cost savings.

 

Q. Will joining the P-NAP replace ISP and corporate connections to the general Internet and their current vendors?

A. No, not in most cases. The P-NAP will accelerate local traffic, and lower its cost but it will not replace being able to reach the rest of the net. However what it will do is relieve the pressure on the long haul connections. If even as much as 15-20% of traffic can be exchanged at the P-NAP then less excess capacity will have to be used for long haul traffic. Also since BGP routing tends to discover the most efficient routing, all traffic flowing from regional providers will tend to be speed along with perhaps 1 to 2 hops knocked off each trip. Naturally clients can purchase transit connections through the NAP.  

 

Q. How do we get started?

A. The answer is we have gotten started. Computers & Tele-Comm, Inc president Graeme Gibson and 4 partners formed the KC-NAP, LLC in Janurary 2004.   Concurrently an agreement was made with the Bryant Building to install a new meet me room (MMR) and make other impovements to help interconnect the many networks located there. A Charter Member Program was initated to quickly have 20 networks join the NAP, so as to increase its viability.

 

Q. Elaborate on that last point?

A. Remember networks become more valuable as the number of endpoints and members rise. Everyone can see the value of the Chicago NAP with 108 networks tied together at 622Mbs. But what is the value of say 6 networks tied together at a 100Mbs level? Clearly it is less, but how do you get started. The first few members will have the greatest risk, what makes it smart for them to be early adopters? Well one means to drive early adoption is a very low cost to entry. Lets say the first 3-5 members for there leadership and motivation to build the thing in the first place are rewarded by lower operational costs or waved entry fees. As the networks grow the wisdom and benefit of joining it will rise, and so to should the costs. If it costs less to be an early adopter the thing will get built much faster, and some of the benefits to the community can start flowing sooner.

 

Q. Are there any other considerations at to how much it costs?

A. Yes, there will have to be mechanisms to differentiate as to service levels and speed. While a Sprint or a MCI may need a 10Gbs connection to the P-NAP a little ISP might do well with a 10Mbs connection. Also there are big differences as to power draws and reliability levels, as well as security levels and add-on services. All of these types of questions will have to be answered as the P-NAP evolves. There also may be valid public service motivations to have reduced costs for educational institutions, libraries, community organizations and government. Many P-NAPs are run, as co-operatives with there own governing bodies and rules of operation, well outside of the traditional business for profit models. In many respects it will be based on the costs of construction and operation.

 

Q. What is the advantage to Charter Members?

A. In order to scale up the numebr of connected networks quickly the KC NAP, LLC has a charter member program that waives most of the up front costs and provides services at reduced rates to the first 20 networks (10 carrer and 10 customer) to connect. These 20 members are rewarded with these lower costs to become early adoptors.

 

Q. How do I get more information?

A. Please contact us via e-mail or call us at 816-842-0606