Some people might think it's been a bad couple of months for reliability. MCI WorldCom Inc. lost a big chunk of its frame relay network for more than a week. AT&T's Internet service was inaccessible for almost a full day. Critical Path's e-mail service has kept some ISP customers away from their e-mail for hours at a time. Sheraton's Web site got hacked and was not taking e-reservations for more than an e-day -- that's seven days in normal time.
However, believe it or not, there is a bright side to these outages. They help company decision makers understand a basic and underlying design principle of the Internet: Reliability is not part of the basic package.
Before the Internet, most telecommunications networks were designed and built to be as bulletproof as possible. That's a great design principle, but it has a drawback: Ultra-reliable systems are expensive systems. High-reliability networks also tend to be slow -- you get superhigh reliability by having elaborate procedures for change and configuration control, administrative checks and balances, and big operations staffs to watch over it all.
The Internet's infrastructure doesn't have that kind of control. It's fast moving, and it's a 99 percent solution, meaning 99 percent of the time, it works incredibly well. And, as we've all discovered, it's dirt cheap. The last X.25 line I ever bought cost US$1,400 per month and a nickel for every packet, at 9.6K bit/sec. Now, I can get a digital subscriber line for $60, run at 256K bit/sec and no one is counting packets. Amazingly cheap.
But these incredibly low costs come with tradeoffs, including reduced reliability. The problem is that some businesses see the lower costs but don't understand that they're giving up something when they save that kind of money.
They think that if they're giving AT&T $19.95 per month for Internet access that they have a reasonable expectation it will work, all the time, flawlessly.
Oh, and no busy signals, either.
That's why I like to see a couple of high-profile outages now and then. It reminds people that the Internet is not 100 percent reliable. As more and more businesses come to depend on the Internet for business-critical communications, they need an occasional reminder of what they're paying for -- and what they're really getting.
The solution is not to throw money at the problem. Buying Internet service from AT&T, as we have learned, is not necessarily getting you anything better than you could get from Joe's Bar & Internet. In fact, the MCI WorldCom and AT&T outages have shown very clearly that big-name companies have no edge when it comes to keeping their networks running.
In the Internet era, reliability is becoming something you have to build, not something you buy. That's hard work, and it requires intelligence, skills and budget. You have your choice. Are you a 99 percent company, or are you a 100 percent company?
(Snyder, a Network World Test Alliance partner, is a senior partner at Opus One, a consultancy in Tucson, Arizona. He can be reached at Joel.Snyder@opus1.com.)