The Internet has some major problems. The big one involves the cost of supporting the amount of content and services being supplied over the Internet infrastructure. And if it's not fixed soon, the expansion of Internet traffic could outrun our ability to pay for it.
Here's why: Since it started, traffic on the Internet has about doubled every year. Remarkably, the router technology that we started with (best-effort packet routing) has supported this huge growth without any basic change except speed improvements resulting from the improvement in semiconductors.
Due to the improvements in fiber technology, the cost of increasing raw bandwidth capacity has been decreasing about as fast as the traffic grows. Fiber, therefore, is no longer the problem. But now that fiber technology has advanced, we have a different problem: routing technology. Internet traffic is now growing much more quickly than the rate at which router cost is decreasing per bit. Traffic is doubling each year, while routers follow the semiconductor trend, dropping in cost per bit by one half every 18 months.
The cost of Internet capacity would therefore double every three years without some key new innovation. The economy could not support this for very long.
Traditional routed IP networks provide reasonable quality by operating with huge overcapacity so the peak usage hardly ever overloads the routers. If a packet router becomes overloaded it seriously damages all the traffic, data, voice, and video. If we don't find a way to keep up with these increasing capacity costs, we'll start to see this damage.
I believe that the solution is flow routing. [Ed. note: Dr. Lawrence Roberts is the founder and CEO of Anagran Inc., a flow-based routing company.] Flow routing has introduced an important innovation that can help alleviate the capacity crunch: Routers do not need to route every packet, only the first packet in a flow. Thus, the inherent cost of these new routers is one third that of packet routers, and they provide an immediate 3:1 capacity increase when they are inserted into the network, eliminating the need to add capacity and cost for a year or two.
Flow router technology can be included at the access point where the overload may occur so that congestion and overload does not damage the traffic; lower priority, large file transfers are throttled back; and interactive voice and video stays protected. This allows the entire network to operate at much higher efficiency, often around 90 percent utilization day and night.
As the technology is further employed, the step function saving is on the order of 9:1 (cost and efficiency). This could extend the time that Internet traffic can continue to double at the current network cost by nine years. At that point, some additional innovation will be needed to keep cost under control or traffic growth will have to slow down.
The Internet's problems are not limited to cost, however. The aging IP technology in the installed base has other challenges.
Quality: Today, video can be easily downloaded just like data, but streaming video only works well if the network has enough overcapacity, with data users kept on a separate network. In many cases (like WiFi, for example), the same is true for voice. We can’t even start to consider many other applications like “telesurgery” -- robotic surgery performed remotely via the Internet -- due to poor video quality as a result of packet loss and delay variance.
There are really two problems to solve here: controlling the huge network load caused by video downloading, and the inherent inability of the current packet router design to support low delay variance, with low-loss streaming media mixed with lots of data traffic. Flow routing could solve both these problems. Based on observing and remembering the state of each ongoing data stream (flow), the router can protect video, voice, and any real-time stream from delay variance and loss.
Security: Security is becoming a serious problem. Although it is partly a computer issue, in large part it is also a network issue, since current networking technologies do not verify who is sending the data. Most known security problems (denial of service, spam, viruses) would be much easier to cope with if the network included three additional functions: authorizing users as they connect to the network; checking the addresses a user claims to be sending from, to insure it is not faked; and detecting traffic anomalies such as denial-of-service attacks.
Authorization is a known technology, but not very useful without source address verification. Source address verification is expensive if required for every packet, but with the advent of flow routing, it can be done once per flow, making it extremely inexpensive. Similarly, detecting traffic anomalies is virtually impossible at the packet level, but quite reasonable with flow routing technology, by simply looking at the flow information.
Thus, with the changes happening in routing technology, we should be able to pinpoint and identify anyone who sends spam or attacks a remote computer, and at least recognize and stop denial-of-service attacks, if not identify the originator. Once security attacks are traceable, law enforcement becomes possible.
Currently, we are expecting the same 40-year-old technology to support not only information exchange like Web browsing and email, but all our real-time traffic such as voice and video. Three basic problems must be overcome to accomplish this: quality, security, and economics. We need to improve packet forwarding design if we are going to fix these problems.
I enjoyed your article. I'd like to know if there's any scientific publication supportting your statement that flow routers increases throughput, reduces delay and improves fairness throughout the entire network.
I am not predicting a collapse of the Internet so please do not be alarmed. What I am suggesting is that without improvements like flow routing in at least the access layer of the Internet, the growing traffic will force a slowdown in growth or an unacceptable increase in cost, a significant increase in delay, and great difficulty in the development of quality voice and video services. However, these problems can and most likely will be avoided by the adoption of new technology.
Anagran flow routers can benefit any medium to large enterprise, as well as any service provider network. If your network is suffering from poor end-to-end performance, or if specific applications do not always seem to get the needed network resources, then you may indeed be a candidate for Anagran flow routers. Speed is not an issue -- all Anagran interface ports run at wire speed.
As far a security goes, Anagran flow routers require a secure username/password to access the user interface, and the product is self-protected against any distributed denial of service (DDOS) attacks.
I’m glad to have sparked your interest in the subject.
No, each router does not have to be enabled for flow routing. Anagran flow routers, once installed at key network aggregation points, make the network much more efficient even if most of the routers in the network are not flow routers. Hence, Anagran flow routers may often serve as an easy “add on” to existing packet-based router networks.
It is an end-to-end solution. Placing Anagran flow routers at key network aggregation points increases throughput, reduces delay, and improves fairness throughout the entire network, including across the core.
The Anagran flow technology can be used to gently meter some traffic in favor of other traffic without demolishing anything. Instead of killing P2P to allow high priority traffic to flourish, for example, the Anagran flow technology allows for “peaceful coexistence” between different traffic types of differing priority. High priority traffic is delivered with required quality of experience, while lower priority traffic continues, albeit at times just a bit slower perhaps.
This technology does not depend on replacing every packet router with a flow router. Anagran flow routers are standard Layer 3 routers. They COMPLEMENT existing packet router networks and have been proven to interoperate with other standard Layer 3 packet routers from multiple vendors. I
n many cases improvements will scale in a linear fashion and in other cases the improvements may be more non-linear. As is always the case, it depends on the network configuration and the overall traffic mix across the network. In all cases, however, an improvement can be expected once flow routers are added to a network.
Interesting as always. With Flow Routing, are we back to logical channels (X.25), DLCIs (FR), VCs (ATM), Tunnels (wireless), Paths (MPLS) and other connection oriented technologies?
Perhaps Flows are smarter than all of the above but they will pose the problem all of these did: scalability. The good-old hop-by-hop magic got us this far partly because it was connection-less.
It really boils down to which is worse: the overhead of creating and maintaining connections or doing packet by packet route table lookups (in addition to creating and maintaing large route tables). Route tables are roughly analogous to area codes and phone numbers which are not maintained by every telephone switch, so a hybrid approach of using switches in the center and routers at the edge as suggested by a reader above might be the next step. If and when we run out of that combination, we can invent some other simplification. The MPLS and pseodo-wire techniques are attempting just that. We don't have enough data to classify them as duds yet.
Do simulations prove if weighted, smart flows improve packet drop performacne? Do we need that improvement at this stage? It would be intersting to hear your thoughts on this subject.
When one reads the messages from this author , there is always a need for reflection of thought before replying. While reading this, take a moment to imagine that you are listening to a favorite beautiful piece of music from a long playing stero record. Comfortable ? Good. You have just set your mind into priority stream data cueing. A little tricky if you have memory interuptus, but pleasurable for how long it lasted. Take a moment to reread Shannon's Law of information theory. List the Agencies of priority for human safety. Put the Q factor (quality) into the equation. Now you have a rudimentary concept of flow streaming.
Needless to say i am a Fanboy of this Author.
Oh CD and DVD can be substituted for Vinyle Record
Cheers all and keep building the last mile. Copper, Coaxial, fiber, waveguide and do not polute the etheric with needless spurious rf. Of course the author is implying Wireless streaming.
I agree fully with the view on optical backbones expressed by awase149. I gave the same prediction in my reply to Drew Lanza's article which I posted yesterday. And, to avoid confusion, that is plain wavelength switching we're talking about here, not the optical packet switching or optical burst switching that you read about in research publications.
Key is that by the time you have a full wavelength of capacity between two edge routers on different ends of the network, you have no reason at all to do routing at the packet level in your backbone. By just switching wavelengths you overcome the limitations of electronics, but you also realize big savings since you can take out a lot of transmitters. I have done a lot of cost modeling on network builds, comparing different technologies, optical/electrical, bitrates, etc, and always found that lasers, modulators and electronics make up a huge part of the Capex. We did this back in 2000, to illustrate the value of optical switching, but of course the traffic levels at that time didn't justify wavelength switching.
When the telecom bubble burst in 2002, there was a lot of stuff on the drawing boards and in demo rooms that was shelved: planar optical switches, MEMS, liquid crystal switches, and all sorts of optical technologies to support high bitrate ultra long haul transmission such as Raman amplification, DCMs, PMDCs and special fiber profiles. Even though many of the companies don't exist anymore, the Intellectual Property must still be out there somewhere, in basements or hidden deep in the assetts of some VC firm (does anyone remember Nortel buying Xros for $3.25 billion?).
When the need arises, all this technology can be easily dusted off and put to use.
Very interesting article!I agree that the Internet core architecture has some major problems.We’ve evolved from the days when best-effort IP routing was adequate to support the primitive applications of email and web surfing.In the early days of the web, the excess bandwidth capacity that was built into the Internet core assured the quality of service and minimized congestion. But today, the explosion in P2P traffic is saturating the core networks. And P2P traffic is projected to more than quadruple over the next five years. Congestion is now a serious problem within the Internet backbone networks.
Your description of “flow routing” is fascinating and I’ll have to read more about it. It sounds similar to IP with DiffServ which is being used to improve QoS in some networks. Or possibly more like MPLS label switching.Does each router have to be enabled for flow routing? Is it an end to end solution or only within the core network?
I have a different approach to improving the efficiency of the Internet. The IP traffic within the core networks could be optimized by moving the routing and grooming of IP traffic out of the core to the edge networks. Employing a high-speed optical backbone with IP over DWDM to eliminate the superfluous layers of protocol and hardware required by SONET and TDM would improve Internet performance and reduce operating costs. The goal is to keep as much traffic as possible in the high-speed optical core network and to use the edge IP networks to optimize and process the IP traffic. Over 80% of all IP traffic within the Internet core is transit traffic that just passes through a router. Therefore, reconfigurable optical add-drop multiplexers (ROADM) could be used to selectively drop IP traffic to edge routers that only process the IP traffic to be terminated. This would reduce hops, latency and permit massive amounts of IP traffic to be carried by the optical core.
Thanks for the great article - - - I will now read about flow routing!
your article on the threatened internet has me somewhat alarmed. We use an ERP business application for our manufacturing facility that runs on the internet. We are in a position to rethink our direction on this. Should I bring this application in house?
The ThinkerNet does not reflect the views of TechWeb. The ThinkerNet is an informal means of communication to members and visitors of the Internet Evolution site. Individual authors are chosen by Internet Evolution to blog. Neither Internet Evolution nor TechWeb assume responsibility for comments, claims, or opinions made by authors and ThinkerNet bloggers. They are no substitute for your own research and should not be relied upon for trading or any other purpose.
There has been widespread discussion lately about the unfairness of the primary protocol we rely on with the Internet – Transmission Control Protocol (TCP) – along with many proposals on how to fix it. Since there are clearly many problems with both slow and unfair service, my question is: Should TCP be overhauled to fix today’s congestion control problem, or does the network itself need fixing?
New York's Metropolitan Transit Authority is conducting a pilot test of digital kiosks to guide subway users to where they want to go more efficiently and at lower cost.
The whole Amazon.reader debate is a double-stupid. It's stupid to think that there's any e-book buyer who doesn't know Amazon's URL, and it was stupider to let ICANN launch the whole free-form TLD initiative to start with.
Edmunds separates customers into segments based on the info it collects on its site and from partners, and uses that to push out custom content, said Brian Baron, director of business analytics for Edmunds.com, at Predictive Analytics Innovation Summit.
The automotive website uses propensity modeling to target ads and customer registration forms, said Brian Baron, director of business analytics for Edmunds.com, at Predictive Analytics Innovation Summit.
Subsidized handsets, rather than locked handsets, should be the focus of regulators. We're not getting good deals, not fostering innovation, and weakening our power as buyers.
The automotive website uses propensity modeling of customer behavior to convert more site visitors into leads, says Brian Baron, director of business analytics, in an interview at the Predictive Analytics Innovation Summit.
Companies need to take advantage of new technologies to simplify interfaces, improve capabilities, and enhance back-office processes. But they can't upgrade their Websites too often.
Cisco's rumored sale of Linksys suggests we may have problem with innovation and profit at the edge of our Internet, and that could be critical to the evolution of many Internet-delivered services.
Many enterprises view high-speed broadband connections as ubiquitous. Yet in about 20 percent of the country, businesses and their employees do not have access to even DSL connections. This shortcoming diminishes enterprises' ability to support their employees.
New York's Metropolitan Transit Authority is conducting a pilot test of digital kiosks to guide subway users to where they want to go more efficiently and at lower cost.
The whole Amazon.reader debate is a double-stupid. It's stupid to think that there's any e-book buyer who doesn't know Amazon's URL, and it was stupider to let ICANN launch the whole free-form TLD initiative to start with.
While NFC's original goal was to enhance mobile commerce applications, it is finding its way into a number of other uses, which is creating both opportunity as well as challenges for IT departments.
Enterprises would like to move to cloud computing but are hesitant because they are concerned about providers’ ability to secure company data. Here are some tips that help to ensure that if breaches occur, the business is not left holding the bag.
Edmunds separates customers into segments based on the info it collects on its site and from partners, and uses that to push out custom content, said Brian Baron, director of business analytics for Edmunds.com, at Predictive Analytics Innovation Summit.
The automotive website uses propensity modeling to target ads and customer registration forms, said Brian Baron, director of business analytics for Edmunds.com, at Predictive Analytics Innovation Summit.
Expert Integrated Systems: Changing the Experience & Economics of IT In this e-book, we take an in-depth look at these expert integrated systems -- what they are, how they work, and how they have the potential to help CIOs achieve dramatic savings while restoring IT's role as business innovator. READ THIS eBOOK
your weekly update of news, analysis, and
opinion from Internet Evolution - FREE! REGISTER HERE
Wanted! Site Moderators Internet Evolution is looking for a handful of readers to help moderate the message boards on our site – as well as engaging in high-IQ conversation with the industry mavens on our thinkerNet blogosphere. The job comes with various perks, bags of kudos, and GIANT bragging rights. Interested?
To save this item to your list of favorite Internet Evolution content so you can find it later in your Profile page, click the "Save It" button next to the item.
M2M: Rise of the Machines? Not Yet David Weldon In the 1970 science fiction thriller Colossus: The Forbin Project, two giant supercomputers from the United States and Soviet Union secretly join forces to take control of the collective nuclear might of the two countries. In the film, the two machines discover each other's existence, communicate back-and-forth, share their collective data, and cut their human creators out of the process. It is the ultimate example of machine-to-machine communications, or M2M. CLICK FOR MORE
M2M: Rise of the Machines? Not Yet David Weldon In the 1970 science fiction thriller Colossus: The Forbin Project, two giant supercomputers from the United States and Soviet Union secretly join forces to take control of the collective nuclear might of the two countries. In the film, the two machines discover each other's existence, communicate back-and-forth, share their collective data, and cut their human creators out of the process. It is the ultimate example of machine-to-machine communications, or M2M. CLICK FOR MORE
Email This
