Slashdot Mirror
Why Are T1 Lines Still Expensive?
badfrog asks: "Over the last 10 years, DSL and cable modem has upped its speed (although in some instances only slightly) and dropped its price. However, the price of a T1 has stayed almost exactly the same. If you had asked me 10 years ago, I would have predicted any geek that wanted to would have fiber or their own T1 line to the house by now. What is with this sad state of affairs that a 'business class' 1.544Mbit connection is hundreds of dollars more than a 6Mbit cable connection? Is it a legitimate case that a high upload rate should increase cost so significantly?"
Why do you even ask this question?
The difference is clear. A T1 guarantees you your bandwidth. Both DSL and Cable do not. You usually get it, but that is only because others only use a fraction of what they are "allowed" to. Look in your TOS, you'll see that they do not guarantee the speeds, they are "averages". So essentialy, your ISP pays for 100Mbps and sells 5000Mbps to 1000 customers (Each 5Mbps, but in reality they get only 0.1Mbps). (Numbers pulled out of my you know what). If everyone would start downloading like crazy at the same time you'd get congestions. The fact is that it's not the bandwith that is interesting with DSL/Cable but the fact that it is always-on.
When DSL started here, it was only 256kbps/64kbps for quite a lot of money. We made the calculation compared to our average ISDN Internet usage (that was per minute) and the price would be the same or slightly higher. Sure, the higher speed was appealing, but the fact that we knew we payed a flat-fee for unlimited interet usage and always-on made it more attractive. That was why we were early adopters, not because it was faster. After all the ISDN 64kbps was plenty of fast back then. It did change our internet habits though: checking the email in business hours was a no-no. We started to check our mail after waking up ;-)
I heared that in Italy you can get a T1 for cheap, but I'm sure it comes with no guarantee.
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
The T1 I purchased in 1996 was $2000 roughly from Sprint. Of that $600 was GTE/Verizon's charge for the loop (2 pair). In 1999 I upgraded to a pair of T1 circuits (bonded) the cost was $2300 total, with $300 per loop to Verizon roughly. Then we split our connection and the Sprint T1 of 2002 cost $975 with $180 of that for the Verizon local loop. So the T1 cost has been dropping. But now the product is not in as much demand. In 2005 when we were moving our ISP to a place where bandwidth was cheap (10-60 USD per megabit/sec depending on the provider we'd chose, we reneted space plus got bandwidth and lost the overhead for the redundant power and HVAC (bundled with the space)), then Sprint offered $655 for a T1.
So T1s have been steadily dropping in price. The local loop charges however are moving upwards as clean copper is getting scarcer in some regions and the install of the box to take fiber and supply a T1 has to be accounted for in the local loop charges now. I have seen deals for $395 all in on the web however. And in the case of Sprint with had a committed information rate of the full T1. The CIR clause will cost a bit on your contract as well.
- Tjp
I am in wallow with my inner money grubbing capitalistic pig. ... Oink!
Is that DS-1s are highly flexible. You can provision a DS-1 a number of different ways. For example you could do 4 channels for data (256k) 12 for voice, and 6 as a private link to another office. Well, the hardware on the back end for all that costs money. That's there regardless of if you want it or not. If not, it's not a DS-1 line. Same reason ISDN is expensive. It's not the exact same and isn't as many channels, but it is a similar technology. Even if all you want an ISDN line for is 128k Internet, you are still getting everything else that one implies, which is quite a lot (a BRI ISDN line is two digital phone lines with all the features).
The old circuit switched digital phone shit is expensive. That's the reason we are moving to all packet switched technologies like VoIP. Much less is needed to run voice, net, video, and VPN over a single link if it is all done over IP. However DS-1 allows all that stuff, but can do it at a lower level. You can break out individual channels and use them for different things.
If that sounds like it's kinda useless, well, it is these days. It's legacy technology more or less. In 50 years, we'll probably see very little if any of it left. Everything will come over an IP connection, and the lower level will be a simply point-to-point with an ISP. However at this point, if you get DS-1, you are paying for all the other shit. Better to find another technology for the physical and datalink.
When I bought my first T1 back in '99, it was about $1200 per month. It was from Shore.Net (now Primus), and it replaced a more expensive 256k circuit from UUnet. In 2001, I bought a second T1 from Sprint for about $950. Nowadays, I buy them for my clients (usually from Speakeasy) for around $400 or less. I'd say that's a pretty big price drop. A dual bonded T1 (as another poster mentioned) is under $800 - well lower than a single T1 cost a few years ago.
Sure, DSL is cheaper, but you get what you pay for to a certain point. Most importantly, ADSL is typically restricted to 768k max upload speed (I can get commercial cable Internet with 1.1 upload around here) unless you get SDSL (much pricier), and then you basically have a T1 without the service guarantees.
-- Josh Turiel
"2. Do not eat iPod Shuffle."
First, a technical detail. T1 lines send their digital signals over lines with high-current, constant DC power. Without a correspondingly high load resistance, the net effect can send hundreds of DC volts through whatever gets plugged in. Don't believe me? Feel free to stop by my data center, lick your finger, and run it across some T1 cross-connects. That's how a lot of old phone company techs look for vacant pairs on trunk lines. It's a lot faster than busting out a multi-tester.
Meanwhile, the equipment that phone companies use for T1 lines is, as someone said, expensive. It's also on a 30-year depreciation cycle. Until that cycle is up, don't expect prices to come down much. Some companies, like MCI, have already gone through a bankruptcy and written off a big chunk of that depreciation, so they might be able to do better, but only if they own the gear they're using. Any telco buying capacity from a baby Bell is going to have to pay (and charge you) the going rate.
Which brings me to the biggest reason for high T1 costs. The price is regulated. T1 lines get billed based on tariff schedules maintained by each state's public utilities commission. That way, small telcos (competitive local exchange carriers, to use the technical term) can theoretically compete with the big guys by selling you comparable service at a comparable price, often by reselling services actually being provided by the baby Bell, with them simply acting as a middle-man.
For the most part, the price isn't the result of supply and demand, or bandwidth guarantees, or idiots who pay more than they should. It's the result of lobbying by the telco industry. And, being regulated by the government, the price is unlikely to ever go down much. The only real competitive pressure on price is coming from MCI and other telcos that are able to give you a heftier discount because of owning their own infrastructure and having a lower cost burden. The tariff schedules are the same for every T1 within any given geography, regardless of who sells it to you, but some telcos can offer bigger discounts off of the tariffed rate if they have lower overhead costs. The effect of that lower cost structure is most noticeable in "lit" buildings, where telcos have large, SONET multiplexer units inside office buildings aggregating all of the data and voice traffic onto fiber and ensuring it stays on their own network rather than a competitor's. In those locations, the equipment is new, with much more capacity at a much lower cost than the gear used for buildouts in the 1990's. There also aren't any third parties involved insisting on a cut of the action.
I work for a telco, and while I'm not privy to the marketing decisions that go in to the whole thing, I can give the reasoning as I've always seen it.
T1s do an amazing job, they are rock solid, and work at distances that DSL simply can't, they have guaranteed bandwidth and service level agreements that involve penalties to the telco if they go down. For companies that truly NEED that connection they're irreplaceable.
All that said, for an awful lot of businesses our DSL packages at 4meg down and 1 meg up are plenty, and a fraction of the cost.
Now for the reasoning, T1s are a royal pain from the telco side of things, they work so well because they use such high powers to make sure that they are heard (close to 300V instead of 52 for telephone) but this causes all sorts of trouble, due to the crosstalk these things put out every T1 line that's installed reduces the number of ADSL customers we can put in the same cable, one T1 line can easily destroy the ability to carry DSL in the same binder group (25 pairs) and over longer distances or with several T1 lines can wipe out the whole cable for DSL. This is a major problem for us, so if we're going to have to work around these sorts of issues, we want it to be worth our while. that doesn't even go in to factors such as the equipment, a DSL modem costs about $50 or less these days, but a T1 "modem" is in the thousands, same deal with the equipment at the other end of the line, then you add the line conditioning that has to be done on longer lines when provisioning a T1, and the list goes on.
DSL is a great product, if you don't absolutely need a T1, then by all means take advantage of the fact that DSL lines are dirt cheap these days.
but when you need a T1 and nothing else will do, don't complain about the cost, it is after all your choice.
T1 lines send digital signals with almost NO current. This is due to the balanced encoding used on the line. There are two primary encodings used in North America (Europe has their own standard): B8ZS and AMI. These encodings ensure that the number of positive signals sent are roughly the same as number of the negative signals sent, resulting in an average DC voltage close to zero. While I don't doubt your anecdote about techs using their fingers to test if a line is live, the signal they experience has more in common with AC than DC.
The electrical specifications of a T1 show that it uses {-12, 0 12}V DC as the signaling alphabet. This is not the "hundreds of DC volts" you claim (maybe you were confused with the POTS system which uses 90V RMS ringer signal).
I don't know much about the politics of the system (I've only designed endpoint equipment and had little interaction with customers), but I know your technical details were rather specious. Do you have any evidence to back your other claims?