The Compaq of ISPs
Wednesday, July 31, 1996 by Dave Winer.
I've written pieces like this before. They usually happen while I'm patiently waiting for my ISP to believe me, and then do something about it. My server, www.scripting.com is back on the air now and appears to be healthy. This piece was written a few hours ago when things weren't going so well.
Network problems! All over the place. As of 8:11AM on 7/31/96, my website is off the air. It's been a comedy of errors. Laurel and Hardy meet the worldwide web. In the month of July, my Internet Service Provider (ISP) has thrown all my cards in the air twice, both times scheduled, and both times without notifying me that they were going to do it, and both times they dealt with me unprofessionally when I contacted them for help getting back on the air.
The latest outage started yesterday at 1:30PM. I sat down to check my email and noticed that my server hadn't responded to a hit in an hour. I checked my email. No connection. I connected to mail.well.com thru a modem connection, avoiding the Internet, signing on the old fashioned way. It worked. The Well is on the air.
I called my ISP, Internex, at 408-327-2200. After five minutes wading thru their voicemail menus, I got a network status report. If you're connecting thru the Palo Alto POP, please be aware that we're off the air for five hours due to planned maintenence.
I knew there was no point in calling them until 5PM. I took a walk. Returned some phone calls. Took a nap. At 5:30, my server started getting some hits, but far too few. The domain name server was still down so I couldn't check my email.
I've seen this before. Their router is broken. I don't know what it looks like from their end, but it's unmistakeable from my end. Been there before. One more time around the loop.
They can't keep the ride smoothe for me. They throw every excuse at the problem, everything other than taking responsibility for keeping my connection constant.
Typically I have two or three ISP emergencies every month. Each time it happens, I have to stop everything. My server is off the air. My customers and partners can't get the software and information they need. DaveNet is silent. My software development halts. So does business development and friendship development. I *use* the net. I've built my professional life around it. My time is expensive. It costs me large amounts of money when I lose the connection. I also lose my perspective.
It's true, in the last couple of years I've learned a lot about networking from all the outages. I've learned how to read a whole new set of tea leaves. Given a few more years of this hell I'll really understand what's going on. But I don't want to understand all the gory details. I want to invent my own gory details.
Every time I get stuck in this position I shrug my shoulders, and say that this is just the way it is. Then, this morning, almost 18 hours after the outage, and my world is *still* not working, I realized that this is the wrong answer. Don't accept this. Much better is possible. Use DaveNet for a good cause. Let's define a business opportunity. A place where the buck stops, where there's professional system management and customer service. You pay a premium for it, but in return for the premium, you get great service.
The way the net works now is no good. Imagine if the same applied to the computers we use. I wouldn't be able to always write software in the language of my choice. Sometimes my scripting software would break, requiring me to use a C compiler to dig out, to rebuild my world. And sometimes the C compiler would break too, forcing me to use an assembler. I don't even own an assembler anymore!
We've come a long way in desktop computers. They usually work. If they don't, hop in the car, write a check and bring home a new computer. You're never down for more than a few hours if you have good backups and if money isn't the object.
This is not true in the ISP world. The norm is outages. Companies growing too fast for their customer's good. A lack of organization on the other end. Lack of good testing procedures. The system is cobbled together. The wires are exposed, they're always tripping over them. Their monitoring software doesn't detect the problems, so customers have to. And often the customers don't have the expertise to explain why the system isn't working. Tech support reps won't treat people like adults (or customers!) unless they speak the jargon. Now I do speak the jargon, so I can see clearly how unacceptable the current situation is.
In 1984 I ran a small software company. We made products for the IBM PC and compatibles, mostly. We were also in the Mac business, but it was a small part of what we were doing. The focus was on PCs. We had a hodgepodge of different brands of computers in the office. We were having the same kinds of problems then that we are having now with ISPs.
Then came Compaq. In my opinion, they commercialized the PC, made it work, and supported the product. When we had a problem, we called them. They solved problems, even though they were growing fast, they were selling something we wanted -- uptime. We could stay focused on what we were doing, we could build on the PC platform, because our PCs worked most of the time, and when they didn't we could easily get them back in order.
We standardized on Compaq. All new machines purchased by the company had to be Compaqs. When we made this choice, the service got even better.
All the battles over standards, Java, ActiveX, etc are moot if we don't have reliable network service. There's an opportunity out there. I first wrote about it in 1994, in a postscript to Mark Stahlman Gets Real. Ironically, you can't read about it today because my server is off the air, and at this moment I have no hope of getting it back on the air. If I struggle enough, hey, it will come back. But they'll knock it out again in a few weeks or days and I'll have to do it all again.
I yearn for the Compaq of ISPs. The no-brainer choice. The one that everyone else is going with. A large capacity net from day one, so I don't have to feel their growing pains. A plan for expanding so that service doesn't degrade when they grow to the next level.
I would help someone bootstrap such a service. But only if they're really serious. There's a huge need. It can't be filled by a fly-by-night approach. This is the one area where we want and need a boring non-innovative player. Make the net work for people who don't want to deal with all the gory details.