|
California Educational Data Processing Association |
The DataBus - Vol. 36, No. 5
|
Two years ago when "Internet" started to reach buzzword status, a few companies tried to position their products as being somehow related to the Internet when they had nothing to do with it. Those companies were forced to back down, and many of them are not actually trying to integrate their products with the Internet.
This year, the scene is different. Internet companies are trying to sell "intranet" as a buzzword, meaning essentially "the Internet, but just within your company." This makes perfect sense, because the technologies that make up the Internet are scaleable--up to accommodate tens of millions more Internet users, and down to handle small companies that want a simple, in-house network.
Nothing, not even a buzzword, is safe from marketing scams, however. Computer companies and magazines are seeing the result of their too-heavy hype of the Internet and are now jumping on the intranet bandwagon, but trying to pervert it at the same time to fit their old LAN-based thinking. What was a reasonably clear market a few months ago has now fallen into the hype morass of the Internet, and it may never make its way out.
There are hundreds of Internet protocols, each specified by the Internet Engineering Task Force. The IETF, which just had its triannual meeting in Montreal at the end of June, promotes the technical side of the Internet by making sure everyone knows what the standards are and where they came from. Want to read the down-and-dirty on the protocols that make up the Web, Internet mail, FTP, and so on? They're all there for free at http://www.ietf.org/.
The IETF has lots of problems, many of which have to do with this high degree of openness. However, it still works well for creating and maintaining the technical underpinnings of the Internet.
By default, then, the IETF also created all the protocols for intranets without really caring about it. Because intranets use the same protocols as the Internet, nothing needed to be changed in order for companies to use Internet protocols on their local networks instead of proprietary LAN protocols like those from Novell and Microsoft. In fact, many large and small companies have been selling intranet software for years (decades, in the case of Sun Microsystems), just without a cute name like "intranet". Instead, they had to call it "TCP/IP".
So, just because you can run HTTP (the main protocol of the Web) on a network that uses something other than TCP/IP, that does not make this network "the Internet" or "an intranet" or anything else. One of the wonderful things about layers of protocols is that you can mix and match them (well, mostly) and they should all work pretty well. But that doesn't give marketing folks the right to say "we run a few Internet protocols on a LAN, so that LAN must be an intranet."
The funny part about all of this is that, because TCP/IP and the other protocols have been openly available to anyone for over a decade, networking companies didn't have to invent their own proprietary protocols: they could have been building intranets for their customers. Had they done this, they might own a large part of the intranet market today. You can imagine how angry this simple fact makes them.
For example, Microsoft is now touting its new intranet strategy. This follows on its recent Internet strategy, and uses most of the same pieces. It's true that you can create parts of an honest intranet with Microsoft tools, but Microsoft neglects to tell you that there are gaping holes in its strategy.
You can run TCP/IP everywhere with Windows NT and 95, and even use fairly robust Web software, all straight out of the box. However, Microsoft still doesn't have an Internet-standard mail system. Its highly touted Exchange server is proprietary, and doesn't use SMTP and POP3 protocols that tens of millions of Internet users use every day. It's as if Microsoft forgot that more people use email than the Web.
In Microsoft's intranet announcements, the press releases blithely danced around this glaring problem by not mentioning email at all. In his speech announcing the new strategy and advertising campaign, Bill Gates said (in a direct quote from Microsoft's transcript of the speech): "We got our Exchange product out and throughout the last couple of years, there's been a lot of neat things going on with Exchange to make it tie into the Internet, to make use HTML to format documents, to make it tie into the kind of exchange SMTP, NNTP that go on with the Web."
Gateways are an intranet marketer's favorite tool because they can say things like, "we don't need to have an SMTP server, we just use our own gateway." They say the same things about HTTP servers: we don't need a real one, we use a gateway. Ah, but if this was only true.
You can think of gateways as translators. Imagine a message in English being translated into Japanese. Even if the translator is extremely skilled at both languages, it is likely that there will be some subtleties that will get lost.
Gateways are, and have always been, the bane of the Internet's existence. Most mail gateways are pretty lame, and are probably the cause of 99% of the lost email on the Internet. That is, it's not the Internet itself that loses mail: it's bad gateways. Even if a gateway is written extremely well, it has to make the assumption that the software that's handling its messages is also written well and is well documented. Yeah, right.
Thus, an intranet system that includes a gateway instead of a real Internet server is not really an intranet. Just as there's no excuse for a LAN to use anything other than TCP/IP, there is usually no good reason for an intranet to use a gateway instead of a real Internet-standard server or client. Today, there are over a dozen SMTP (mail) servers, and over fifty HTTP (Web) servers. For a company to say, "Oh, it's too hard to write our own server," or even worse, "Our gateway will work better than an Internet server," is like saying, "We don't care enough to do the real thing."
Now that Microsoft gives away a reasonable set of some TCP/IP software in Windows 95 (although it is still a real pain to configure), you would think that they would take the last step and go all the way with helping people with intranets. They haven't yet, and that's letting Netscape get a fair amount of market share selling their intranet solutions to people Microsoft could be serving.
By this time next year, this could all be moot, with people setting up intranets by getting some of what they need from Microsoft and the rest from other vendors. Of course, the big loser will be Novell, whose bread and butter comes from non-open LAN software. This is not to say Novell will go out of business soon: there will always be a legacy market. But as more and more people want to make their in-house systems work the same as the Internet, intranets will quickly start to outnumber old-style LANs, giving everyone many more choices from a variety of vendors.
Return to August-September index