Is the Internet backbone running out of bandwidth?

[seerongo]

I keep running into reports that within a year or so, the current system will be unable to sustain the rate of growth of bandwidth demand. Projections range from "the internet will become little more than a toy" (a paraphrase), to predictions of brownouts or other limitations. It seems that the investment in new technology required to prevent this is so expensive that it is unlikely to happen any time soon.

Having been around for awhile, I'm very skeptical about doomsday reports like this by now. It's hard for me to imagine that the industry would allow the system to deteriorate to any extent. Also, most of the blogs and reports you get from googling, seem to all derive from one report from IBM last year. It just is all too reminiscent of the Y2K scare. I spent two years trying to reassure my customers about that one (at least for my products) and it wasn't easy.

Any professional insights on this? There should be a good consensus on this by now, but all I see is controversy.

 

[Pattonias]

I imagine that if this was to happen then the new technology that would be required to prevent/fix the problem would cease to be too expensive.

 

[DavidSnider]

In any case there is a ton of room to be made by simply updating a few wasteful internet protocols.

 

[Evo]

I keep running into reports that within a year or so, the current system will be unable to sustain the rate of growth of bandwidth demand. Projections range from "the internet will become little more than a toy" (a paraphrase), to predictions of brownouts or other limitations. It seems that the investment in new technology required to prevent this is so expensive that it is unlikely to happen any time soon.

Having been around for awhile, I'm very skeptical about doomsday reports like this by now. It's hard for me to imagine that the industry would allow the system to deteriorate to any extent. Also, most of the blogs and reports you get from googling, seem to all derive from one report from IBM last year. It just is all too reminiscent of the Y2K scare. I spent two years trying to reassure my customers about that one (at least for my products) and it wasn't easy.

Any professional insights on this? There should be a good consensus on this by now, but all I see is controversy.

First, there is no single "internet backbone". That is a term that we in the field use to refer to the thousands of companies that have agreements to hand off traffic to each other. Now, major telecom carriers have larger pieces than smaller companies, and in rural areas there can be bottlenecks where a small local provider will have to go to a public peering point to hand off and receive traffic.

I don't know the current stats, but there is a ton of dark fiber available. This is fiber optic cable that does not have the equipment connected that sends signals across the fiber. The equipment is very expensive. Existing fiber can carry more and more traffic simply by adding equipment capable of increasing capacity. All major carriers engineer their networks to keep at least 50% extra unused capacity in their networks at all times.

Unfortunately there may not be capacity in the central office to accept new connections, and cost studies and engineering requirements for build outs, delivery of equipment, etc... have an impact on adding capacity for smaller ISP's that wish to buy circuits (capacity).

I'm at work and that's a reall quick and dirty 2 cent version of "why the internet is not running out of bandwidth". Now how much money companies are willing to keep investing in expanding capacity is driven by the economy.

 

[seerongo]

I should clarify a few things here. First, I mentioned a report from IBM. That should have read AT&T. Here is a news report referencing that:
http://news.cnet.com/2100-1034_3-6237715.html

Thanks Evo for correcting my misuse of the term backbone. I probably should have said "internet" and let it go at that.


DavidSnider said:

In any case there is a ton of room to be made by simply updating a few wasteful internet protocols.

Are you referring to IPv4 vs. IPv6? If so, could you expand on that?
There seems to be a bit of controversy about the impact of that on overall capacity. It would seem that hardware limitations are the potentially the most imminent bottleneck, then more efficient protocols may become necessary.

From Pattonias:

I imagine that if this was to happen then the new technology that would be required to prevent/fix the problem would cease to be too expensive.

That's my take too.

I am still interested in the potential bottlenecks.

 

[Evo]

I should clarify a few things here. First, I mentioned a report from IBM. That should have read AT&T. Here is a news report referencing that:
http://news.cnet.com/2100-1034_3-6237715.html

Basically, this is what I said in my post. More capacity = more equipment = more money. Major backbone providers, I work for one, have to be able to turn a profit, so unlimited internet is becoming a thing of the past. I can see the internet going back to a "pay as you go" service. The more you use, the more it's going to cost. On one hand, companies keep putting out more devices encouraging people to use more internet, then complain when customers actually use the service. Most people use a modest amount of bandwidth, but some people really abuse it, and those are the ones I see getting targeted first. I don't see any way around it. It's not magic, the networks cost money to maintain. No one has a "right" to use as much as they want. Like the article says, the companies that form the "internet" are numerous privately owned for profit companies.

I sell backbone access to ISP's, it's what I do for a living. I worked for AT&T designing and selling data networks from the late 70's. I jumped ship after 30 years of handling data networks for AT&T to work for a competitor.

I am still interested in the potential bottlenecks.

What kind? Peering points? Central office ports? Backhaul?

A recent incident was a client of mine, a medium size telephone company in the south wanted more bandwidth in order to provide internet service to it's customers. They wanted to meet me at one of my POPs in another state for geographical redundancy. They provided their own circuit to near the state line where their authority ended, then rented a circuit from a small third party carrier to the city where the POP they wanted to connect to me was located. The problem was that this small carrier did not have the necessary clearance to come into our POP two miles down the road, so my engineer knew of another local fiber company that did have the ability to cross-connect to us in our POP and agreed to do the last two miles of fiber to interconnect the two buildings. Great, but we didn't have any spare gigE ports and I had to get a job approved to get the hardware installed so we could do the final cross connect, basically a drop from one floor to the next. Yes, I am over simplyfying and not using telecom jargon so that it is easier to picture how it happens, I hope.

 

[seerongo]

Evo: Thanks for the example of what goes on behind the scenes. As for "Pay as you go" service, I guess you would see a metered scheme per GB or something, as we pay for other utilities such as electric. Remember when nuclear was going to be so cheap it wouldn't be metered? Actually, as a customer, I think that makes perfect sense and personally, I would probably come out ahead in the long run anyway. It would probably make you think twice every time you hit the "Enter" button, and maybe that would be the biggest advantage of all...

 

[seb7]

The backbone is not a problem. New lines are constantly being laid down which carry several terabytes of information. These are used also for other types of communications as well as for the internet.

The problem being experience at the moment (and getting worse), are that ISPs are being pushed to offer more bandwidth to their customers when there own switching equipment is already running at maximum bandwidth.

To fix this, they are using equipment which gives priority to different types of internet data. They are enforcing their 'fair use' policy to limit and cut off heaver users. They really don't like heavy users. Most users no longer actually get the bandwidth they pay for.

What ISPs should be doing is buying more bandwidth from the backbone, replacing all copper lines with fibre, and use faster fibre switching routers. Its a costly upgrade which is a little too expensive for most ISPs.

---

On a similar note, most comms companies seem to only be upgrading to meet its current demand. Its obvious the internet bandwidth requirement is never going stop growing. So why they are not thinking further ahead I don't know.

 

[Evo]

On a similar note, most comms companies seem to only be upgrading to meet its current demand. Its obvious the internet bandwidth requirement is never going stop growing. So why they are not thinking further ahead I don't know.

Cash flow. It's not good business sense to have your money tied up in unused facilities.

 

[seb7]

"ISP offers 1 terabit broadband to homes." - How quickly would that sell out!

 

[dreiter]

I hate the idea of tiered internet. That's like someone buying a cable package and the company saying they can only watch TV for X number of hours a month. If you pay for those channels, you should be able to watch them as much as you want. Similarly, if you pay for 10mbit, you should be able to use that as much as you want. Perhaps the cost per megabit would need to go up up in order to satisfy the economic strain on the companies, but they certainly don't seem to be hurting for money right now...