Slashdot Mirror
Myths about Internet growth
An anonymous reader writes "An article in The Economist outlines WorldCom's role in starting the myth that Internet traffic doubles every 100 days. This helped inflate the telecoms bubble."
← Back to Stories (view on slashdot.org)
According to this, there was about 1 gbps of internet traffic in 1995.
If this doubled every 100 days, there would be 50,000 terabits per second of internet traffic today. There's actually less than one terabit/sec of traffic.
By 2010, we could expect more bits per second of internet traffic than there are atoms in the universe.
A legparnasom tele van angolnaval.
It usually turns out that assault means "felt uncomfortable and/or threatened in an ecounter with the opposite sex." How many of us haven't felt uncomfortable? I'm surprised the statistic is a mere 1/3
Actually, the study that showed this (often quoted) statistic said it was 1 in 4... and it wasn't just "felt uncomfortable"... one of the questions was "Have you ever had sex, then regretted it the next morning?" *bing* If the participant answered yes, then the "study" said she had been sexually assaulted.
An interesting side note to this is that one of the questions was "do you believe you have been the victim of a sexual assault?"... the response rate of this was 9%..
So even though only 9% of the women polled thought they had been sexually assaulted, the people doing the study still claim 25%..
from Jakob Nielsen
http://www.useit.com/alertbox/9509.html
traffic at (T=0) = B
:)
delta time(years) = DT
traffic at DT = B*2^(DT*365.24/100)
hmmm... lets say there was 1 KB of traffic per hour in 1990 (ridiculously low...) B=1, DT=12.
2^(43.8) = 1.53e13 KB = 15.3 petabytes of traffic per hour today. Somehow I doubt it...
A big part of the confusion that comes from these kinds of statistics is failure to understand the problem well. There are many parts to the equation of Internet growth and these statistics are a gross oversimplification to anyone who's in the telecom industry.
A simple example is the difference between access networks and core networks. Access networks are those such as phone lines or DSL to the home. Core networks are those such as UUNET. Companies buy big pipes from those guys. Both types of networks need to be in place and operating before Internet access will grow. If one slows down, the other slows down because they are positively correlated (I hope that's the right term).
So, you've got the access networks (ISPs and RBOCs for example) preaching their growth numbers and core networks (UUNET, Global Crossing and AT&T for example) preaching their numbers and no one looks at the big picture.
Taking this example to the Internet boom, where we are today is that core networks are way overbuilt and access networks suck eggs. The ISPs just don't have the subscriber base to support them buying more bandwidth from the core network providers.
(Keep in mind, this is on the service provider side of the equation. There is a whole other side that's the enterprise side of the equation, where big companies buy bandwidth from access and/or core network providers to connect their people together. It's a big part of the market and is almost the only thing keeping these network providers afloat these days until the service provider market returns.)
"Money often costs too much" -- Emerson
can any of you point to any of these studies to support your claims?
The "one in four" study, done by Mark Koss was commissioned in 1985 by Ms. magazine; It was published (in Ms. magazine) in 1988.
To the best of my knowledge, the Ms./Koss Study is not anywhere on the web; however, it (along with a few others) is covered very well in the following:
http://www.leaderu.com/real/ri9502/sommers.html
But is there really a difference in the cost of providing bandwidth (other than hardware, which is still a fixed cost) if I want a low end ISDN line or an OC3 pipe? I would liken it to cable TV. Somehow I manage to now get digital cable with a few hundred more channels than its anolog predecessor for the same price. The cable company didn't really have to do all that much other than give me a new cable box (which I rent from them).
Sadly, not true. First, there are big differences in the cost of the customer prem equipment between different networking technologies. You can buy a cable modem for $80 or less in bulk - a SONET box capable of supporting an OC-3 link will cost you ~$15,000. That being said, if you want internet access, your local access provider (be it the phone, cable, or whatever company) is going to have to purchase a connection to the backbone, and that costs money too (figure $40k for a 155Mbps link). Secondly, on the cable front, going from a few analog to a lot of digital channels required billions of dollars in capital expenditures for the cable companies. First of all, they had to upgrade all the amplifiers and passive components in their networks, and introduce a lot of new elements, since the higher frequencies used for digital channels attenuate over shorter distances, and hence need to be re-amplified more often. In many cases, they had to replace the actual physical cable as well.
My phone line has been there for years. Other than a $50 line cleaning kit, what is really the increase in cost for me to get DSL? Other than equiptment that the telco buys to provide DSL. If they buy it to provide access to one user, what is the increase in cost when you add another 300 users? I understand that hubs and routers have physical limitations, but it just seems that we are getting porked.
The equipment on the other end of your DSL connection (called a DSL Access Multiplexer, or DSLAM), doesn't come cheap, and the unit's cost scales pretty close to 1:1 with the number of ports it has - net/net, the cost does rise with the number of customers. Beyond that, the telephone companies had to make some pretty significant adjustments in their network architectures (pushing fiber down the network, for example), to make DSL available on a widespread basis.
I might back up the above by comparing it to wireless (cell) phones vs traditional land lines. Don't the cell division make hand over fist compared to the land line division? I mean, they put up a tower that can service thousands of people. No cables (from telco to your phone) and thus significantly less service persons and cost per customer. It makes me wonder if my inflated cost for use of a cell phone (in my opinion) is there to offset the money-sucking land line division... Shouldn't cell phone service be only a small % of the (my) cost of wire-based telephone service in my house?
Actually, the cellular service providers are LOSING money hand over fist. First off, building those towers isn't cheap, esp. when you consider the legal aspects of getting access to the sites, permitting, etc. Second, they have to buy landline connections from the towers to their networks, and that is usually slow and expensive. Third, there's a limited number of users you can support from a given tower - spectrum isn't endless. This doesn't even begin to consider the costs of obtaining the spectrum itself - look at the billions of $ that the European wireless carriers paid for 3G licenses there. That being said, it's certainly cheaper, if no network exists, to serve an area with cellular than build a physical phone network. A lot of phone companies in the developing world are doing just that (Telmex in rural Mexico, for example). The real costs of cellular, on an operating basis, are operating - marketing, customer service, etc. The industry still has really severe bad debt problems, and customer churn is high, so getting and keeping customers sucks up a huge amount of the revenue pie.
Overall, are we getting hosed? Basically, no, I don't think so.
Internet traffic doubles every 100 days? I never heard that myth. They must have a bad marketing department.