Google
 

 

 

 

Stash the Cash!

The Intranet Gang is Here

Francis Vale

Grand theft is about to happen! Your budget for client server development is going to be swiped Worse, the corporate crown jewels, your data assets, are also about to be snuck out the door. And the go between that fences your precious assets across the entire planet is flaky at best, and horrendously insecure at worst. Fortunately, we already know who the soon-to-be alleged perp is: It's the notorious IntraNet Gang.

The Internet has suddenly gone from something 'out there', to being The Intranet, a virtual corporate WAN 'in here'. Intranet connectivity holds forth the promise of easy information access, anytime, from anywhere within the organization. Such simplified, universal access was the organizational paradigm that propelled client server into a multi-billion dollar industry.

If the Intranet was just a virtual WAN pipeline, then the transition from client server computing to Intranet computing would be relatively painless. But the Intranet is also the Internet, and as such, along with it comes a tremendous amount of social, political, and technical baggage. Which helps explain why it's not enough for the IntraNet Gang to merely swipe your client server development money. On the gang's backroom agenda are plans to snatch millions of dollars from your present and future IT/IS budgets. This break-in to your glass house is going to cost you plenty.

How can this happen? Who is letting this motley crew into your company? And what can you do about it?

First, the bad news: The IntraNet Gang cannot be stopped. Nothing you do will get rid of them. More bad news: There is more than one gang at work here. Your IS/IT crew better hunker down, or it will get bushwhacked in the hyperbole crossfire. So where's the good news? Maybe it will be easier if we just line the perps up against the wall. We'll let you pick out the Intranet culprit that's going to cost you millions.

One IntraNet Gang calls itself Microsoft. Its tactic of choice for erasing turf competitors is its operating system stranglehold. Its rival gang across town calls itself Netscape. This Netscape bunch hangs out with another tough crowd, called Sun Micro. Both Netscape and Sun share the same gang mascot, known as Java. Once Java sinks its teeth into you, it may prove next to impossible to shake the persistent critter off.

Java started out life as a software system for embedded microprocessors. It's code name was 'Oak'. But Sun Microsystems finally came to its senses, and recognized its primary business was selling UNIX workstations and servers, and not appliances, fax machines, and cell phones. Hence, the Java marketing shift.

Java is meant only for 32 bit systems. Depending upon one's perspective, this may or not be a bad thing. But it does beg the question of what will be done about the millions of installed 16 bit MS Windows systems. Despite Microsoft's most ardent wishes, the whole world will not be going to 32 bit Windows95 or NT over night.

So, if your organization has a large installed base of Win 3.x systems, and you plan to gradually phase them out over several years in favor of Win95 or NT, then you have some immediate incompatibility problems. (It has been reported that IBM plans to create a Java implementation for Win 3.x. But such a 16 bit implementation will be at applet odds with all the other 32 bit Java systems.) The obvious solution: immediately chuck your entire installed base of PC operating systems and applications, along with most of the hardware. If this massive upgrade is not in your 1997 budget, then it should be -- if you want Java as an organizational Intranet standard.

When you buy those machines, make allowances not only for the demands of their new, nicely fat operating systems, like NT, but also for Java. Java is a multi-tiered system. The crux of Java is a standardized virtual machine (VM) which uniformly rationalizes all the elements of the underlying operating system. Hence, anywhere a Java applet finds a Java VM, it sets up and runs. No Java code recompilation is required. For the heterogeneous Internet, this looks like great news. Write a Java applet once, and ship it off anywhere on the 'Net. The Java bytecode interpreter will execute the Java applet, be it on a PC, Unix system, Macintosh, etc. This is a great notion, and is the principal driving force behind Java. But Java is not truly platform neutral. For example, if you use threads for a SPARC/OS, and special GUI/platform features, then Java application independence becomes problematic.

Also, if you have ever run an OS emulator, say a Win/Tel emulator on a Macintosh, then you already know Java needs CPU power. Lots of it. The VM approach, especially for a 32 bit VM like Java's, sucks up cycles like no tomorrow. If you add this Java overhead to your already burdensome Win95/NT requirements, then you know that the minimum amount of RAM for your new systems must be at least 32MB, and preferably 64MB. Moreover, that new CPU had better be Intel's fastest, i.e., most expensive, chip. If you don't spring for the biggest and the fastest PC, then get ready to watch the MS hourglass ad infinitum. This is a prospect guaranteed to make your suddenly idled users very unhappy. Bottom line: the multimillion dollar PC/OS/application upgrade program you originally planned to occur over 24 to 36 months has just shrunk to twelve months; and it's going to cost you hundreds, or maybe thousands, more per PC unit -- if you want your Intranet Java now.

This high CPU overhead is what prompted Sun to recently introduce a line of embedded microprocessors ("Pico-Jav") and Java operating system ("Java-OS") specifically optimized for executing Java applets. In effect, the whole multi-billion embedded microprocessor industry has just been told that its products don't cut the Java mustard. This must come as some news to SGI/MIPS, who make one of the world's fastest and most powerful line of embedded RISC CPUs.

The Java language can be thought of as a kind of C++ dialect, with some Objective C extensions. It is an object-oriented system. For those who like the idea of creating reusable class libraries, this is great. But for all those other users who have discovered that continually creating new class libraries, or revising old ones, sounds good theoretically, but in practice falls apart rather quickly, this may not be such good news. Most IS/IT shops have probably had the experience of gung ho object programmer initiates creating scads of class libraries, only to discover down the road that real world production realities forced them back to their old C sinning, unstructured, ways.

Furthermore, how many C++ knowledgeable programmers do you have in your IS/IT shop? And do you want to take them away from other demanding projects to work on Java? No question, you will need more of these hard to find C++ stars. Plan on spending another $60K to $85K for each of these new Java code warriors. And that's before figuring in annual overhead.

Finally, the not-often-discussed big size of Java applets will make onerous demands on network bandwidth when downloading. Their large size will come as a rude shock to those 'Net surfers who already complain about those 50KB+ web graphics. The easy answer to this problem is that higher bandwidth is coming, so why worry? But anecdotal evidence indicates that even a 128KBs ISDN line is too slow for downloading Java applets.

So even assuming you can get ISDN service (which is still problematic, even in places like downtown Boston), and the TELCO has not suddenly skyrocketed its ISDN rates (a price hike phenomenon now occurring all over the U.S.), it is still not the probable answer for moving Java applets around your Intranet. But multi-megabit ATM networks will do Java quite nicely. How many ATM LANs do you have installed currently? So, to your list of new expensive PCs, and additional C++ programmers, you can now add several hundred to several thousand dollars per ATM port pop. And don't forget, all your bridges, routers, and gateways will have to be 'Java' upgraded as well. Perhaps even your current wiring plant has to go.

For all of the above reasons, it should now be obvious why Java has swept the computer industry by storm. Can you think of any other recent technology that will cost your organization so much money for so many different kinds of upgrades?

But despite Java's performance/platform problems, and relative immaturity, it has already become a de facto standard. MIS really likes the server-side aspects of Java, as all the Java apps reside on easily managed servers. (Marimba's Castanet-Java, however, can reside on the client's disc for off-line viewing; unlike Sun's Java, which currently can not.)

The Java bandwagon is rolling, and every vendor worth its salt is riding on it. The question is, will your IS/IT Department be on this bandwagon, or crushed under its wheels?

But assuming your newly Java-ized budget is approved, and you manage to stay on top, rather than being rolled over, then your IS/IT group can also begin to think about using VRML, Virtual Reality Modeling Language. Netscape, Silicon Graphics, and Sun have jointly announced plans to VRML enhance Java. (SGI's OpenGL 3D modeling technology formed the basis for VRML.) Moreover, as part of its ever increasing buying spree, Netscape has just bought Paper Software, the developer of WebFX, one of the most popular Netscape browser plug-ins that support VRML. To its credit, WebFX's plug-in offers seamless integration of Internet 'virtual worlds' with Netscape.

Essentially, VRML transforms your 2D static Web page images into a 3D space that can be flown over, walked through, or otherwise dynamically 'viewed' from a wide variety of angles. Like HTML, ASCII is the basis for VRML codes. Thus, you can create VRML web applications with your text editor, just as you do for text-based web pages. Likewise, just as you can dynamically link text across different web sites, you can also tie distributed VRML-based virtual worlds together, and even to other text. However, given the obvious difference in data types, that's about as far as the similarities go between HTML and VRML. This means yet another learning curve if you want to fully to exploit the VRML universe.

Numerous VRML developer kits are coming on the market, so you can avoid the tedium of using a simple text editor. For example, Caligari (Mountain View, CA) has come out with several authoring tools that feature VRML support. Virtus WalkThrough Pro is another such tool. InterVista also offers VRML software development kits. And Paragraph International supplies a state of the art VRML authoring tool., offering simple point and click 3D world construction.

Microsoft also supports VRML in its Internet Explorer browser via Virtual Explorer. And Microsoft has its RealityLab 3D engine for fast virtual worlds rendering (RealityLab is based on SGI's Open GL). Both InterVista and Caligari are suppliers to Microsoft's VRML efforts. Caligari's software will be showing up in Microsoft's Internet (Blackbird) authoring systems. Intervista's VRML work is to be found in MS Explorer. Naturally, SGI's Web Force systems provide built-in, easy to use VRML tools., should you go the UNIX way. All these vendors' VRML-generated codes are supposedly platform neutral. But like Netscape's HTML 'extensions' marketing ploy, you can reasonably expect that a number of incompatible VRML enhancements will soon be appearing.

Apart from VRML code incompatibility, two other technical issues will also raise their ugly heads, both of which can increase your already groaning Intranet budget. The bit mapped images that VRML worlds use for textures eat bandwidth for breakfast. VRML files also consist of long -- very long -- three dimensional coordinates that also have to be downloaded. Add them all up, and you can find yourself hard pressed to create VRML files smaller than 100KB in size. In this instance, at least, there is some hope for bandwidth salvation. VRML files lend themselves quite nicely to data compression, so some of these nasty resource hogging effects can be overcome (but you can also expect various and incompatible compression schemes to emerge).

However, when uncompressed, these VRML files will still chew up CPU cycles. (Remember, VRML started out life on powerful and expensive 3D UNIX workstations.) So you have some design tradeoffs to make when crunching an Intranet virtual world. With VRML, it is recommended that you use no more than 10,000 polygons when creating virtual worlds for Pentium class machines. Also, to keep file size down, you should stick with basic colors, with very little surface texturing. But 3D computing power is getting cheaper all the time, so these constraints may slowly ease.

Regardless, you still have some costly Intranet choices if you decide to give your aging COBOL database a VRML facelift: 1) Buy all your Intranet users something like a Digital NT Alpha workstation, for around $6,000 (about $2K more than the going rate for a high end PC). Digital is claiming that its new Alpha PC NT machines are up to 60 times faster at 3D rendering than a Win/Tel machine. Or upgrade all your PCs with 3D graphics accelerator cards. High quality 3D cards range in price from $600 to almost $2,000. Once again, there is no free lunch, as you also have to factor in support and installation costs of those cards across the enterprise.

Finally, designing 3D virtual worlds is not a walk through the park for ordinary programmers, who are usually not trained to think in 3D interactive dimensions. Designing a proper virtual world requires a specialized talent, like good graphics design.

To put VRML into perspective, ask yourself this simple question: Do I really need a virtual world to accomplish my IntraNet mission? 95% of the time, the answer will probably be no.

But if you are in the 5% that say yes, then in addition to your new, big, fast PCs (with 3D accelerator boards), new 32 bit operating systems and applications, C++ Java specialists, ATM ports, and communications/wiring systems overhaul, you must now add a 3D virtual worlds team to your IS/IT Intranet shopping list. (How much are we up to now?)

Fortunately, there are some easier, less costly alternatives if you don't want to bite into the whole Java/VRML enchilada. One good possibility is Macromedia's new Shockwave software. Essentially, Shockwave takes Macromedia Director movies, compresses them up to 60% (via Macromedia's Afterburner utility) and sends them down the 'Net. 3D graphics, audio, and all the rich multimedia effects that Director can be used to create, are now available via one software development package.

To play a 'Shocked' movie, you need Macromedia's no cost, run time Shockwave plug-in (Netscape 2.0 only; and it also requires 5MB of disc space). Significantly, unlike 32 bit-only Java, Shockwave is available for 16 bit and 32 bit PC systems, including MS DOS, MS Windows 3.x, as well as Win95 (and also Macintosh, with SGI and Sun support on the way). The Director movie plays as soon as it is downloaded into system memory, i.e., the video/audio works in 'streaming' mode.

But this still does not get around the network bandwidth issue. Even though the movie files are compressed up to 60%, 40% of a multimegabyte Director file is still a lot of bytes to move down the 'Net. You can expect to wait a few minutes on a 28.8 modem. Moreover, Shockwave is not in the same performance category as other such streaming systems, like RealAudio or VDOLive (the latter being a Netscape plug-in that offers real-time 'Net audio and video playback). Shockwave is still far away from delivering such real time multimedia performance to the Web. But it does deliver the promised goods.

To create a Shockwave movie, you will need Director 4.0 for Windows or Macintosh. The next Rev of Director, 5.0, will also be offering its own version of Netscape 3rd party plug-ins, called Xtras. And finally, Director 5.0 will overcome one of 4.0 greatest weaknesses: text creation and manipulation.

At $800, Director is not exactly inexpensive. But when you consider the costs of hiring specialized Java developers, VRML designers, etc. that $800 begins to look like a bargain. If you had intended to use Java for adding colorful animation, moving logos, training or tutorial programs, creating interactive kiosks, or even multiplayer games, then Shockwave is probably a better route. But you will still need a developer on your staff who can specialize in Director. However, such a person will be easier to find, and much less expensive to pay, than a Java guru.

At this point, it is worthwhile to consider the new software market dynamics which the Internet has wrought. At one time, software was produced and marketed according to classic business principals. A company produced and priced a product based on its perceived market demand. Free market competition then decided if that product/price was appropriate. But in the Internet era, everything has been turned upside down.

Netscape helped pioneer the notion of 'free beta' Internet software. The gambit was to give things away at no cost, get them well promoted, and then expect revenues from the sale of auxiliary products. In Netscape's case, the revenue cookie was its Web server. For Macromedia, it's Director sales they are after. In both example cases, the intention is to get your users to download free software, have them play around with it, and then have them demand that your company support it. Thus, the business dynamic is no longer market demand push, with product pull-through. In the new Internet model, it is vendor push, with market pull-through. One side effect of this new model has been to rewrite the rules of the software industry. If you have a hot new, barely baked, product, then you just put it up on the 'Net, promote it like crazy, and then have millions of confused users and hapless support people create a market for you.

But for one key industry player, this new Internet game is old hat. Microsoft has long been a master at the vendor push/market pull scheme. With its operating system franchise at nearly 85%, it has mastered the fine art of the software 'giveaway'. E.g., was it really necessary to have a market 'beta' test of almost a million copies of Win95? It should therefore come as no surprise that Microsoft has announced that it will be 'giving away' both its Explorer browser and its server software. In the case of NT 4.0, the Web server will be bundled in, 'for free.' Of course, Microsoft does nothing 'for free.'

Microsoft has now embarked on a seismic organizational change in an attempt to reposition itself for maximum Internet market advantage. Unlike IBM, which idly sat by while the new PC industry rewrote its business game book, Microsoft is aggressively reacting to the Internet's market shaping force.

Microsoft has a number of marketing arrows in its Internet quiver. Each one, no matter how 'free' it may be, will cost you. The first and most obvious freebie, will be intrinsic, Internet support within all mainline MS applications. For example, if your organization uses Excel as part of its decision making structure, and an Intranet is your new internal communications vehicle, then Internet-enhanced Excel will become a must have. The same goes for MS Word, etc. But Microsoft will not be upgrading its Win 3.x applications with such Internet utility. These features will only be found in Win95/NT. So once again, like Java, you must immediately plan to upgrade all your 16 bit systems if you are to reap the benefits of an MS-style Intranet.

The next MS giveaway we have already mentioned: the 'free' NT web servers. Expect Microsoft to closely couple in all of its Backoffice features with its NT Web server. This will also require your buying a complete set of Backoffice upgrades. And most especially, plan on Microsoft tightly coupling its SQL Server together with NT Web server. Tying the two together makes perfect sense, as you will want seamless integration between data requests hitting your web server and your databases. Expect MS Web/SQL Server to be another costly upgrade.

Finally, there is the 'free' MS browser, Explorer. It used to be that in a side by side product match-up, the Netscape browser was clearly the features winner, and readily bested Explorer in overall functionality. However, given Microsoft's grind 'em down track record, one could have expected this product features disparity to soon disappear. And it mostly did, with Explorer 3.0.

But the Browser Wars are now over, anyway, as browsers sold as stand-alone products are now passe. Navigator is now merged into Netscape's new Communicator Suite, which includes E mail, Navigator 4.0, HTML editor, conferencing tool, and groupware clients. With Communicator, Netscape suddenly has two big market competitors: IBM/Lotus Notes, and Microsoft. MS Explorer is also being absorbed into MS Windows. Indeed, watch how MS Explorer and the Win95/NT desktop GUI become ever more functionally intertwined. Microsoft has a well established track record of integrating 'add-ons' within its operating systems. This is especially obvious as Explorer becomes the "container"/GUI for Office 97. In Office 97, in place navigation becomes the logical analog to in-place activation.

Leveraging this tight integration, Microsoft is introducing an interesting add-on product, dubbed Microsoft Merchant Server. Microsoft bills Merchant Server as the "complete Internet commerce solution for users and business." Merchant Server offers electronic shopping extensions for clients, which include encryption and payment along with a local order cache. Server application support includes links to product information, order database, as well as support for legacy order entry and billing systems.

Another key Backoffoce component closely tied to the Internet Information Server is Exchange Server 4.0. This brings messaging and groupware support to the Web. Enhanced Internet Support for Exchange Server includes the ability to publish public folders as live web pages, a Network News Transfer Protocol (NNTP) connector that allows the reading and posting of Usenet news articles to public folders, and support of POP.3 clients.

Look too, for Microsoft to rapidly enhance 'Blackbird', now reincarnated as the Internet Studio Virtual Publishing System. (Blackbird was originally intended for the MS Network, before the Internet stole Microsoft's attentions.) Studio has many things going for it, including substantial ease of use in creating content for Web pages. It features built-in wavelet compression, progressive rendering, plug-in ActiveX editors, ActiveX controls, and even Netscape extensions. Think of Studio as a DTP system for Web content.

ActiveX is the next logical step in the evolution of Microsoft's object technology, OLE, which no longer carries a version number nor remains an acronym for object linking and embedding. Essentially ActiveX provides the means to connect Win32 and OLE-enabled applications to the Internet.

All of the familiar OLE and OCX (OLE custom control) components are there in ActiveX, along with the underlying Component Object Model (COM), which has finally become distributed via support DEC-compliant Remote Procedure Calls (RPCs), not to mention support for URL monikers.

As a result, developers can readily leverage their existing skill sets and tools as they move out from the confines of a desktop-centric world. Existing 32-bit COM applications, such as EXEs generated with Visual Basics 4.0, and OLE applications generated with MFC automatically work with Distributed COM. But what appears as a clear boon to developers is what jaded pundits label as the ActiveX "marketecture."

So. what's the scoop? Is anything really new with ActiveX?

Perhaps most fundamentally, Internet functions have become just another element to be structured inside an OLE-driven container applications. Only now the containers encompass not only desktop applications, such as MS Office and Visio, but also Web browsers.

While distributed COM is the basis for ActiveX components, these components are nothing more than COM objects. As traditional COM objects, they expose their features to collections of methods in strongly typed interfaces.

Each object has one ("IUknown") or more interfaces which enable objects to have discoverable features for knowledgeable clients. And clients still call these methods as indirect function calls through a table ("vtable") of interface pointers. What's new is "free threading" for ActiveX applications, which should increase the scalability of object servers. D/COM/RPC under WindowsNT 4.0 manages a pool of threads that come up through vtables when calls arrive. What's more, the range of ActiveX components, which are made up of documents, controls, and scripting engines, is expanding.

ActiveX documents ("DocObjects") are an extension to OLE's in-place activation, a feature introduced in OLE 2. In-place activation allows a container application to activate an embedded object without showing a separate server window. With the addition of several new DocObject interfaces, Internet Explorer now becomes what can be described as a container application for ActiveX applications.

Unlike DocObjects, which occupy the entire client area and have to handle menu merging and toolbars, OLE controls are small autonomous inside-out objects (inside-out objects are activated the moment the container is opened) usually implemented as in-process servers via DLLs. Furthermore, OLE controls can be implemented programmatically by using automation to expose objects to OLE scripting written by the user. It can be fairly stated that in the Web/Internet world, more useful software has been written by many more people using scripting languages like Perl, or TCL than in C or C++. With ActiveX, scripting languages (engines) are also expanding from the traditional Visual basic Script to include PerlScript, and JavaScript.

Sun has countered Microsoft's ActiveX push with Java Beans, another OS-level component technology, that lets Java apps work with ActiveX and OpenDoc. IBM has also endorsed Java Beans. Distributed Java uses Sun's "Joe", based on the OMG ORB (Object Management Group/Object Request Broker) -- But the Java/Internet Inter-ORB (IIORB) is not yet standardized. Each ORB needs its own specific Java applet. You cannot mix and match multi-vendor ORBs.

Netscape also endorses the OMG/ORB model. Furthermore, Netscape says it will support ActiveX as native browser code, and not simply as a plug-in. But its Java IIORB is not based on Sun's Joe IIORB. Instead, Netscape uses the ORB from a small company by the name of Visigenic. If Sun and Visigenic don't come to terms, then Netscape might become incompatible with Sun, IBM, IIORBs. (Note: Oracle is about to announce an "open" IIORB API between its systems and Netscape's.)

Microsoft has countered the OMG/ORB efforts by "donating" parts of ActiveX to the Open Group, a standards organization. But don't be fooled. ActiveX is still a MS proprietary technology as it is fundamental to Microsoft's operating systems. While OLE and COM are relatively free of platform dependencies, ActiveX controls are not. ActiveX code is specific to the Intel architecture and the Win32 API; neither of which exist on UNIX or the Macintosh. So if an ActiveX control wants to draw pixels on a screen, it must access the Windows graphics APIs. But these APIs do not exist on the Mac, nor on UNIX. System services use APIs that are all different from one OS to another. Indeed, if the Win32 API was truly open, Microsoft would be helping Sun Micro to run Windows applications on UNIX systems. This is most unlikely! Furthermore, ActiveX is pre-compiled code that is loaded up across the Net. How do you anticipate heterogeneous systems? Also, unlike the Java VM, ActiveX has no built-in security system.

But don't be taken in by Sun's open systems mantra, either. For like the MS ActiveX/Win32 APIs, the Java APIs are also controlled by one company: Sun Micro. Sun Micro also owns the JavaOS, and--unlike MS--also sells CPUs (Pico-Java).

In short, this is all just marketing gang warfare Neither Active-X, nor Java is an "Open Standard."

Historically, Microsoft has won in the marketplace as much by the ineptitude of its rivals as by its own business brilliance. But as we have seen, neither Netscape and Sun have shown no such slovenly behavior. For its part, Netscape has aggressively branched out and broadened its product line, either via in-house development or, more likely, via acquisitions. For example, Netscape bought Collabra, the groupware maker, in its bid to strengthen its hand against Lotus notes. Netscape also bought Aurum Software, a maker of software that merges Web-generated sales leads into a company's internal sales and marketing systems.

One big item still missing from Netscape's application shelf, however, is database tools. Here are at, least, there is strong support from Next, and its new WebObjects. WebObjects is a comprehensive enterprise solution for creating and deploying object based systems across the corporate Intranet. Via this Next software, organizations have a set of tools to easily link their web servers into databases from Oracle, Sybase, and Informix. Next is also rightly famous for making such distributed object creation relatively quick and painless. But it should also be noted that WebObjects are completely web browser and server independent, so all of its quick to build and deploy benefits are also applicable to Win95/NT Explorer clients, and NT web servers. However, if you don't commit totally to Redmond, then you won't be able to exploit all of the company's Intranet product benefits including WebObjects.

Thus, the proper bottom line query becomes not how many budget dollars you have to spend, but rather: Do you feel comfortable betting the enterprise/Intranet farm on Microsoft? The real cost of 'free' MS software is not in the manifold upgrades you will have to buy. Rather, the true price to your organization is having to become an all Microsoft-Intranet shop. It is Microsoft's ultimate goal to transform the Internet into the MS-branded Intranet. In the Microsoft world view, such 'Net market branding gives it all the benefits of having its own SNA or DECNet, but without any of the network support and development hassles.

Numerous companies have successfully made such single vendor bets in the past. For example, many users placed their entire MIS corporate fortunes in the hands of IBM/SNA. Others, with equal good fortune, went down the all Digital VAX VMS/DECNet path. This single vendor/technology model worked quite successfully, for a long period of time. So it might reasonably be expected that it will also apply to the MS enterprise Intranet. But then again, maybe not.

The rate of technological change is far faster today than it was in the 60's, 70's, and 80's. Back then, going all DEC or all IBM meant product/technology lifetimes measured in years. Today, these lifetimes are measured in months. Can any one small company be expected to embrace it all, and enhance it all, at this high velocity of change? (And yes, even the multi-billion dollar Microsoft is still small. IBM sells twice as much software per year than the Redmondians).

So, does one go with cross-platform UNIX-based servers, and machine independent Netscape clients; or with MS NT servers, MS Explorer, and the soon to be MS-branded Intranet? UNIX Web servers are proven, robust scaleable solutions. NT is rapidly, but not yet, at that level. And the jury is still out on whether Microsoft or Netscape emerges as the ultimate market share/features winner. And can any one company assume complete market responsibility for something as vast and unwieldy as the Internet? For that is how uses will perceive Microsoft's new corporate responsibility, should the MS Intranet brand succeed in its market positioning. This attempted MS-branding may prove one bridge too far for Microsoft.

A cynic might say it is the vendors who will win these Internet marketing battles, and it is users who will lose. Unless, of course, users specify Internet and Web standard multimedia protocols; strict HTML "standards" adherence; cross platform availability for all componentware (ActiveX, JavaBeans); machine independent applications; cross platform, easy to administer, Net management tools; and robust security. But with such a rapidly moving market, and such ferocious vendor wars, even these standardization guidelines may prove to be useless. The moral: spend your Internet/Intranet dollars very carefully, and astutely calculate the true return on your Internet dollars. Just bear in mind, however, that in this new era of IntraNet Gang Warfare, using traditional ROI techniques may not have much meaning any more. Many of the Intranet vendors and technologies you choose today will most likely be dead/obsolete within 12 months.

If you are the CIO, you had better brace yourself for some Al Capone times. And if you haven't already done so, now is the time to start educating upper management on what it means to walk down these new IS/IT Mean Streets.

So double-lock the MIS bank vault! The IntraNet Gang is coming.


Copyright 1996, Francis Vale, All Rights Reserved

21st, The VXM Network, http://www.vxm.com

s