Slashdot Mirror

grammar nazi wrote

Whenever the "barriers to entry" (i.e. initial fixed costs) of running a business are low, then many people go into business and compete with each other. Due to this competition (competition==good thing), profits become lower and lower. Finally, the razor thin profit margins means that only a large corporation can make any substantial profits, driving the smaller companies out of business.

Ummm, no. Or, at least, not necessarily. There are substantial diseconomies of scale associated with business, and in a service business like the ISP business, there are few or none of the economies of scale that are associated with manufacturing businesses.

For example, the total tech support payroll at the small mom-and-pop (I'm the "pop") ISP that I own is smaller than SWBell's payroll for tech support managers. Those tech support managers don't provide tech support, but they do add to the cost of providing tech support. Therefore, tech support is easier to afford for the smaller ISP than for a company like SBIS.

Of course, the razor thin margins are there and cause failures, but the companies that tend to fail are those that borrowed a bunch of money, (or sold a bunch of stock,) often to buy up smaller companies, and have discovered that they can't make enough beyond their overhead to make the interest payments. In this case, the small ISP's are the ones being bought, not the ones doing failing

Anyway, two things have been obvious to me for some time. First, providing access to the Internet to individuals is not going to be a license to print money, like people were thinking it was going to be five years ago. As you say, the barriers to entry are too low for that to happen. So, the successful ISP will need to focus on other areas to make most of their profits. What Brokersys tries to do is to have the dial-up stuff to pay for its share of the fixed cost, but we make most of our profits on other things. Second, many people will pay a premium to buy their Internet from a smaller company. It is, therefore, a mistake for a small ISP to try to price themselves like the big discounters.

markbark wrote

Would you charger per bit.... per byte?
Would you charge for useage or perhaps throttle back those who used too much bandwidth?

Being in the ISP business, I have a slightly different perspsective on the matter, although I intend to read the full paper, once I get a chance.

The reason that we've never offered a metered service, even though a few people have asked for one in order to reduce their bills, is because we've never considered the work needed to keep track of users usage for billing purposes to be worth the effort.

Two answer your question, there are two broad schemes used. The first is the peak bandwidth scheme (used by those who sell "burstable bandwidth") where you pay for the peak data rate you use usually with some averaging and time dependance. (In the most recent deal proposed to me it wasn't clear to me what the penalties were for exceeding the base rate.)

The second is to simply charge by the bit, possibly with a certain amount provided at the base rate. For example, for 1.5 MB DSL service, you might be given, say, 50 gig per month (which corresponds to a utilization of about 10% of your line's capacity) at the base rate (maybe $10 per month for the bandwidth only.) plus, say, $1 per gig after that. I wouldn't meter outbound traffic at all. There's no point. I also wouldn't meter the traffic from your premise to my equipment, so you can check your mail as often as you'd like or load the Web page that shows your current month's usage without fear that that will put you over your quota.

I'd suspect that even heavy-duty Web surfers and email addicts would have trouble getting anywhere near the base rate, and I'd offer fixed-rate service (maybe $20 per month) for the Napster users and guys who browse the binaries newsgroups.

In my opinion, the key to customer acceptance of this mechanism is twofold. First, you need to offer a fixed rate for those people who want it. As the article points out, many people will pay substantially higher for a fixed-rate service than a variable-rate. Second, you have to make it easy for people to know what their usage rate is.

One reason why people who have cell phones will pay extra for a large flat-rate plan instead of choosing a metered rate plan which might actually cost them less money is uncertainty in their usage. When starting out, most people don't have any idea how much they're going to use their phone. Once they now, then it isn't worth the bother to make the change. Take away that ignorance any fewer people will make that choice. Make it easier to switch and people will.

I will tell you that although I worked out this scheme in some detail, it's not likely to be put in place. That's because the largest part of the cost of providing the service doesn't have anything to do with the upstream bandwidth, which is all this scheme meters.

symbolic wrote:

For starters, there's one key element missing from the open source movement...since no one owns the software, no one is accountable. I doubt seriously that the mainstream corporate world is going to embrace something for which there is no accountability.

The standard reaction to this objection is: There is no accountability with proprietary software, so why are you requiring it of free software? The reason that it is the standard reaction is because it is so true. The saying "no one ever got fired for buying Microsoft" (which, of course, used to have "IBM" where the "Microsoft" is) just illustrates the point that most companies aren't even going to hold their own employees responsible for bad decisions as long as they're the right bad decisions. If that's the case, why should those businesses hold their vendors responsible for those same bad decisions?

Furthermore, what are the consequences for those vendors? No large software vendor cares what a single customer thinks about their product because, no matter how big the customer is, it represents only a tiny amount of revenue. Whine about W2K all you want, but if you're a Microsoft shop, you'll buy it in the end because it's what Microsoft has to sell you.

symbolic also wrote:

Here's something else to ponder...open source is the way it is now because of the way things are now. Microsoft is the big bad enemy, and Linux is the underdog. Who is to say that the dynamics of open source won't change dramatically if this balance shifts? Right now, open source is a cause. If it hits the mainstream, I posit that it will join the endless clatter of every other mundane process, and since the open source "cause" is no longer an issue, another will have to be identified to take its place.

I haven't checked, but I'd wager that there's at least one person who has written a comment to this article that goes something like "software is just a tool, people should use what makes sense instead of insisting on free software". The problem with that statement is that it implies that everyone using free software is using it despite the fact that it offers less value. Free software users are, from this perspective, cutting of their noses to spite their faces. You repeat that slander here.

However, my view is that software is just a tool and that everyone needs to choose what makes sense for them and I choose to use Linux as my operating system. This choice is not made because free software is some sort of cause celebre, but because it works well in the applications that I have and you can't beat the price, even if you factor in the cost of self-support. I simply don't have the resources to support two operating systems.

In fact, and to more directly address your point, in Support is Not the Issue that the shift to free software is the result of fundamental economic forces similar to those that brought the ISV's to their predominant place in computing. Even if Microsoft goes away completely, free software needs no antagonist to continue to be the best solution for a wide variety of circumstances.

chris.bitmead wrote

Is there some reason they had to hack pppd just because they are covering 400000 sq km? I don't see other ISPs hacking pppd.

That's because most ISP's have enough funding to purchase access equipment rather than rolling their own. For what it's worth, I hacked pppd similarly for Brokersys about four years ago. For a while, there all our dial-ins were on custom-built Linux-based terminal servers. I did all the pppd hacking in a couple of weeks. The idle timer took longer, but worked better than the one in the Ascend Maxen we use now.

However, I decided that as soon as we could get funding for such things that it was better to buy someone else's solution. The issues involved were the desire to purchase ISDN PRI's for the telephone service along with the desire to offer ISDN dial-in access to customers along with the realization that I had better things to do than maintain a really crufty pppd hack.

It also turns out that, if your revenue level supports it, you can get really good financing deals from the equipment manufacturers. For the last three years, Cisco's been trying to convince us to lease some AS-5200's, which we probably will never do as we're deemphasizing the dial-in stuff.

Anonymous Coward Wrote:

By "server room power supply vendor" I meant "the company we bought the *huge* server-room power-supply conditioner / UPS from". Sure, it's supposed to take care of things, and there must be some kind of warranty, but 1) you'd have to get them to pay up, and for a multi-million dollar claim there's gonna be serious paperwork, to say the least, 2) then you have to find the hardware, and you can't just buy a Starfire at CompUSA, 3) you have to rebuild the whole cluster

This is, in a word, stupid. If you have to worry about power problems destroying your power conditioner, then you've bought the wrong power conditioner. The setups I've seen (and I run a wimpy one-lung little ISP so I don't have any of that kind of serious hardware, oh and if you go, be sure to check out our y2k policy it's at the bottom of the page) the only danger the power system can be to the protected computer is if a power pole is driven through it by a tornado or something.

From my perspective, either what you've got works, or it doesn't. If it works, then you've got no reason to go off-line, and every reason not to. If it doesn't work, then you shouldn't be up at all if you're afraid of the risks because the risks tonight aren't going to be any greater than the risks any other night.

Tomorrow, on the other hand, is a different story. Think about it: 90% of all the cops of the face of the earth will be sleeping off the 36 hours of duty they have tonight, but why would anybody worry? After all it's not Y2K.

Anonymous Coward Wrote:

By "server room power supply vendor" I meant "the company we bought the *huge* server-room power-supply conditioner / UPS from". Sure, it's supposed to take care of things, and there must be some kind of warranty, but 1) you'd have to get them to pay up, and for a multi-million dollar claim there's gonna be serious paperwork, to say the least, 2) then you have to find the hardware, and you can't just buy a Starfire at CompUSA, 3) you have to rebuild the whole cluster

This is, in a word, stupid. If you have to worry about power problems destroying your power conditioner, then you've bought the wrong power conditioner. The setups I've seen (and I run a wimpy one-lung little ISP so I don't have any of that kind of serious hardware, oh and if you go, be sure to check out our y2k policy it's at the bottom of the page) the only danger the power system can be to the protected computer is if a power pole is driven through it by a tornado or something.

From my perspective, either what you've got works, or it doesn't. If it works, then you've got no reason to go off-line, and every reason not to. If it doesn't work, then you shouldn't be up at all if you're afraid of the risks because the risks tonight aren't going to be any greater than the risks any other night.

Tomorrow, on the other hand, is a different story. Think about it: 90% of all the cops of the face of the earth will be sleeping off the 36 hours of duty they have tonight, but why would anybody worry? After all it's not Y2K.

From the old Fidonet (newsgroup? can't remember what they were called then) C discussion area.

IIRC, they're still on the web (though not as comprehensive as I remember them). http://www.brokersys.com/snippets/

they'd be a good start for something, in any case. There are lots of goodies (though some really old, irrelevant stuff too) in there.

wesmills wrote:

You should come live in the Dallas/Fort Worth area. Sure, it's not the prettiest place in the world, and we have our share of problems, but we also don't have people who make "only" $35,000 a year living on public housing

Or Houston (where I live just fine in a 2100 SF 4BR ranch I lease for $695/month) or San Antonio or Austin. (I hear Austin is a nice town to live in if you can stand to be that close to the legislature. Fortunately, it only meets every other year, and they're done for a while.) For that matter, you could go to Oklahoma City, Des Moines, or Jackson, MS. As for Internet access, Houston has flat-rate ISDN and DSL (cable modems may or may not be coming soon, depending on who you ask,) at decent prices. (Can you say DS-1 speeds for well under $100/month? I thought you could.)

However, the decision of where to put a company is not made by the workers. Not that it should make any difference. I know that if I were to start a company, I wouldn't do it in Silicon Valley because the costs are too high. I would think that it's cheaper (and probably easier) to hire people in CA and move them to someplace cheaper than it is to pay them a living wage near SFO. The thing is, I suspect that few of the bosses read /. (It's something to think about when you want to start YOUR company, though. Your nest egg goes that much farther if you don't have to squander it all on luxuries like housing.)

To be sure, you can't "go water skiing in the morning and snow skiing in the afternoon" in Houston, TX like someone once told me they did in LA, but if you're supposed to be working 80 hours/week (or doing some other ridiculous things for the benefits of your boss and the detriment of yourself) there's no time or energy left to go skiing at all.

I certainly don't wish ill on my colleagues in the Bay area, but if you haven't put down roots, yet, and you're sick of paying vastly too much for your housing, food, and whatever, you might pick up some out-of-town newspapers and do some hunting, or just go to The Houston Job finder and look and see what's there, or, rather, what's here.

Brokersys (the company I started) isn't hiring, though, so please don't send any resumes.