Something "Fundamentally Different" Heads Your Way

BW's technology columnist, Stephen Wildstrom, discusses the impact of Microsoft's upcoming software release, plus other IT trends

Following are highlights from a presentation by BusinessWeek Technology & You columnist Stephen H. Wildstrom and his answers to audience questions afterward at the CeBIT technology conference in New York City on May 27:

Stephen H. Wildstrom: I was originally billed here as taking the long view, which would have given me the opportunity to talk about DNA computers and quantum computing. And it would have been really easy because I could have said anything and nobody, including me, would have had any idea whether I was right or wrong for a very long time.

But instead I thought it would be a lot more useful to focus on a much nearer term future. So I'm going to be talking about things that are happening, or that I expect to happen, in a time frame starting the third quarter of this year and maybe ending around 2010.

And the first thing that's going to happen that will have a real impact on IT managers is going to happen very soon. And that is Microsoft's release of Service Pack 2 for Windows XP.

Service Pack 2 is fundamentally different from anything Microsoft has ever done. Like any service pack, it contains a roll-up of all the bug fixes and security patches that Microsoft has released since Service Pack 1, which was about a year ago. But the big thing about SP2 is that to a considerable extent, it replaces the existing security model of Windows XP. It changes a great number of installation defaults to a much more secure configuration, which isn't hard considering that the default configuration of XP is essentially hopeless. And it changes a number of security internals to get what Microsoft promises will be a much more secure computing environment than we have ever had with a Microsoft desktop operating system.

Within the enterprise, deploying it won't be too bad because it's managed. Use SMS or whatever and you can deploy what you want and set your defaults according to policy.

Where it's really going to be an issue is when people start deploying it on their own machines that are occasionally connected to your networks. For one thing, it's going to break applications. They're trying to keep that to a minimum but applications that operate at a very low level, especially those that integrate with the security system -- leading candidates being anti-virus software and virtual private networks -- are likely to break.

The second thing is it's going to cause sort of an interesting support problem because it turns on a new firewall, which is a much better than the old Internet Connection Firewall. Your users will find that when they run an app the first time, it's going to ask for permission to allow the app to connect to the network. This is going to confuse them. And it's going to create a lot of burden on help desks. So it's one you're going to have to prepare for.

The much more interesting challenge is Longhorn, which I imagine everybody knows is Microsoft's code name for the next version of Windows, which is due whenever. It's hard to be very definitive about Longhorn because it seems to be morphing every day. It's clearly a project that is in some difficulty. So it is not clear when it's going to ship, and it's not clear what it's going to be. Based on what Microsoft has been saying about it, it will be at least as important a change for client machines as was the move from Windows 98 to Windows 2000.

But the most important thing in it is it is for the first time really in the history of Windows, it is a total change in the file system. NTFS, the basic Windows file system, remains there at the bottom. But on top of it the file system of Longhorn will be a true relational database. You will have a database database with a bunch of fields. The actual contents of the file will be one of those fields, a large binary object field. The rest of the fields, which will vary with the type of the file, will be metadata.So for the first time we're going to have a file system, which actually knows something about the files. This is conceptually a huge difference from a file system which knew where files were located, maybe when they were accessed and who accessed them but essentially knew nothing about their contents.

It's a big change and they are clearly having some trouble pulling it off. They've already pulled back on some of their more ambitious plans to integrate WinFS with the .Net frameworkin a way that would basically make your desktop and the network completely indistinguishable. They can't pull that off in the time frame of the client release. They still hope to do it eventually.

The plan had been to release the first Longhorn beta at the Microsoft Professional Developers Conference this fall. That has now slipped to the first half of next year. They're still talking about a release to manufacturing in the first half of 2006. Frankly, I think that's very unlikely because history tells us that for an operating system release of this magnitude, the beta period is a minimum of 18 months not a little under a year.

So I think at the very best they might be able to ship something by the end of 2006. The caveat on that is that if they slide their ambitions back enough, they can ship it earlier. But they'll be shipping a lot less.

The server version, the replacement for Window Server 2003, won't ship before 2007. And you won't really be able to get the full benefit of Longhorn until the server version is out. And the really full .Net version of the WinFS file system will probably not be there until the end of the decade. The third really big trend is the explosion of voice-over IP telephony, which I am increasingly coming to think is the next big thing information technology.

A lot of companies are already doing what amounts to local voice over IP. They're either replacing or supplementing their existing telephone gear with IP keysets on desktops, and maybe using IP circuits, more often leased than public internet, to link branch offices.

That's where it starts. But voice over IP is going a lot further than that. It strikes me as very unlikely that anyone replaces a PBX or moves into a new office and puts in a new PBX is going to put in an old-fashioned TDM proprietary PBX. They're going to do voice over IP because it's all upside.

One of the nice upsides is that you're no longer locked into proprietary systems. You can use a Cisco switch, and you can buy Zultys keysets. It's all standards based. There's a fairly new protocol called the Service Initiation Protocol that is making this all possible. And any SIP-speaking keyset can speak to any SIP-speaking PBX. It's a very, very different world from the closed, proprietary world of telephony that we're used to.

But the advantages go beyond that. Say you have a telecommuting employee. With voice over IP it becomes very simple to give a telecommuting employee an extension at home with all of the features they'd have on a keyset in the office. No trick to it at all. No paying your local telephone company's huge fees to install a foreign extension in a different area code. You just do it. You don't need any telco intervention.

People are going to figure out ways to use this kind of flexibility and low-cost power in telephony that I think we have a very hard time even imagining today. It's a paradigm-changing approach because today your switch vendor and the local telephone company are in command. With voice over IP, you are in charge.

The question is what's going to happen to the local telephone companies in the process? Nimble is not a word that comes to mind when describing the Verizons and SBCs of the world. But their life will depend on their nimbleness over the next couple of years. They're going to have to find a way to either get out in front of this parade or get run over by it. And right now, I'm betting on them getting run over.

In the long run, the public switched telephone network as we know it is going to go away, to be replaced by a pure IP network. This finally will erase what's left of the line within companies between data and voice telephony. Voice just becomes another kind of data.

Okay, number four, wireless. Most of you are already grappling with the problems of deploying and managing wireless services for people in the field. You are going to have to support multiple technologies and multiple carriers. One technology or one carrier is never going to be available in all places. Supporting multiple air links, multiple technologies, and multiple vendors is going to make life easier for the guys in the field but harder for the managers back home.

So we have a bunch of contending technologies. Wi-Fi is terrific for use on campuses and in public spaces like hotels and convention centers. It is not very good as a wide area network. And the whole economics of the public hot spot business is dubious. Those providers are struggling badly.

The dramatic thing that's happening is that the cellular-type networks are getting vastly better at delivering data at good speeds and sometimes at reasonable costs. On the CDMA networks, which is the technology used by Verizon Wireless and Sprint, you can generally get speeds equivalent to dial-up just about everywhere in the United States wirelessly. Verizon has deployed an EVDO network, a data only network. Currently, it's only up and running in Washington, DC, and San Diego County. But they have been quietly upgrading their base stations in all major markets. They'll have a number of major markets, I think including New York, online by the end of this year. And they promise to have all all major markets will lit by the end of next year. EVDO gives speeds of 300 to 500 kilobits per second. So you're talking about slow but true broadband speed on a wireless network.

Things are moving a lot slower in the GSM world, which geographically is most of the world and about half of the United States with AT&T, Cingular and T-Mobile. Their current data network is GPRS, which gives you a little less than dial-up type speed. And their migration path is a lot slower. They will eventually get to the same kind of speeds that we have in the CDMA world but it's going to take longer as they go first to an interim technology called EDGE, and over the longer term a very high-speed technology called UMTS.

The wild card in this mix is something called WiMax or 802.16, which works a lot like Wi-Fi but over considerably longer ranges and at considerably higher data rates. One big problem with WiMax at the moment is that its major technology proponents, Intel and Motorola, have radically different views of what it is. Motorola is promoting it primarily as a fixed-point wireless service to be a last-mile rival to cable and DSL. Intel has a more ambitious vision of WiMax as a metropolitan area rival to technologies like EVDO and UMTS. One big question about that is where they're going to find the spectrum. And it's also not entirely clear that technology will really work that way. But I've also learned not to bet against Intel when they decide to put a lot of money into a project. And that's what they're doing. So WiMAX could be really interesting.In addition to supporting all sorts of different wireless networks and technologies, you're going to have to support all kinds of different wireless devices. In most large enterprises you have varied users, varied needs, varied usage patterns.

Delivery of wireless email remains the most important wireless data application. BlackBerry has been a clear leader. Blackberry's is evolving from being a pure hardware company to making its service available on a bunch of different platforms. Palm is also strong in the space with the Trio 600, whose sales have been limited only by the ability to get the devices produced.

Microsoft is coming on strong with both Pocket PC and Smart Phone formats. Pocket PC still suffers from not having a keyboard, which in the email world and in many other applications. That's going to change fairly soon.

And the final contender--not terribly important so far in the US but very important in the rest of the world, especially Europe--is Symbian. It's a consortium in which the majority stake is now owned by Nokia. And Symbian devices have proven quite popular in Europe. You're going to have to support them all or at least a large subset of them, and this is going to be a challenge.

My fifth and last trend is Linux on a desktop. Now, I suspect just about everyone who does IT management has gotten at least some Linux deployed in the server space. Linux's penetration in desktops has so far been very small. There are a number of reasons for this. The main is that Linux desktop user interfaces have not been very polished.

That's changing in a fairly significant way, mainly because of the increasing involvement of some very major players in the Linux desktop space. We have Sun with their Java Desktop System; Novell, which bought Ximian, and they have the Ximian desktop. There's Xandros. There's a bunch of other providers.

Another major force that's going to drive much better Linux on the desktop is IBM's Linux desktop initiative. A few months ago, Sam Pamisano wrote a memo to all IBM managers setting a goal of getting Linux on half of their internal desktops by the end of next year. That's a lot of desktops. And it's a lot of industry leadership.If IBM can actually make it work, it's going to create a much, much larger market for Linux on the enterprise desktop. Because if IBM can do it internally, IBM Global Services GS can deploy it externally and it's going to be a big deal.

The big remaining problem with Linux on the desktop is the application space. Application development has lagged way, way, way behind operating system development. At the moment the only places you could deploy Linux on a desktop is in very controlled environment where you don't need many apps. Obviously you're fine on browsers. You're fine on email clients. Ximian even has a pretty decent Exchange client for their desktop. For productivity apps, there's Star Office from Sun or it's open source version, Open Office. Whether they can really be deployed as an alternative to Microsoft Office depends on the type of user and the demands. In some circumstances, they can.

The problem, of course, is most companies have a huge number of custom apps, and most of those custom apps have been written for Windows. And everyone has to think very hard about whether it is worth the cost and trouble of porting those to Linux. It's an economic calculation every company has to make it for itself.So those I think are the five really big issues facing IT managers in the near- to medium-term future. Of course there are little things like budgets and funding and stuff like that. But I don't really know anything about that so I wasn't going to get into it.

And I think we have some time left. So I'll be happy to take some questions.

Q: Which programming languages that you feel will be dominant in the next few years?

Stephen H. Wildstrom:

I think that the big two will continue to be C++ and Java. C-sharp is obviously getting some traction and will get more as Dot Net matures and deploys more. But I think it will be primarily remain a C++ and Java world.

Q: For voice-over IP to really take off, don't last mile issues have to be addressed? Does cable have enough capacity?

Stephen H. Wildstrom:

Okay, well it depends whether we're talking residential voice-over IP or business voice-over IP. For business voice-over IP, we don't really have last mile issues. For the home, it depends. I've had pretty good experience -- actually quite good experience -- running voice-over IP over an ordinary residential DSL line. Voice doesn't really demand very much bandwidth. One of the reasons that voice-over IP has a significant cost advantage over conventional telephony is conventional telephony is phenomenally wasteful of bandwidth, The bigger issue is quality of service and latency. Clearly for this to really work, we got to get QoS standards built into the Internet. It's coming but it's not here which is why most companies that are using voice-over IP to link offices are still doing it more over leased lines than over the public internet.

Q: I'd like to get your opinion on where do you think going to happen with J2EE [Java 2 Enterprise Edition] and .Net.

Stephen H. Wildstrom:

But clearly we've got both out there and we've got bitter rivalries. We've got totally incompatible systems both trying to do mostly the same things. I long term we have to find a way to get these two worlds to come together. Don't ask me how that's going to happen.As to which one dominates, I don't think either does. I think they just sort of split up the market. Maybe it's 50/50. Maybe it's 40/60, 60/40 -- I don't know. But I think for the foreseeable future you're going to just have both approaches out there.

Q: IBM's Linux Desktop Initiative, are they developing their own software or are they using something like Novell?

Stephen H. Wildstrom:

IBM did make a significant investment in Novell at the end of last year and they and Novell have gotten very close. They're locked together in this legal fight with SCO. So my guess is that IBM would be most likely to go with Simian. Someone from IBM [in the audience] is nodding and saying yes. So, that's the answer: Ximian.

Q: What's your thought on the significance of China for Linux?

Stephen H. Wildstrom:

The OS market, especially the desktop market, in China is fascinating. Obviously Linux's initial attraction to the Chinese was that it was free. It was free and legal as opposed to Windows, which was sort of free and illegal. Microsoft estimates that at best 5% of the copies of Windows that are running in China are licensed. Microsoft has for a long time turned a mostly blind eye to this situation

Recently, Microsoft has been changing its song a little and coming up with easier licensing terms which basically boils down to lower prices -- much, much lower prices -- for developing countries.

That's a hard business model to sustain for very long when you're dealing in something as fungible and easily transported as software. It's hard to sell for different prices in different markets.

Microsoft either has to price the product at a price that developing countries including China can afford to pay or they're going to lose those markets. And on the other hand, they have to worry a lot about these low prices creating serious problems in their existing markets in developed countries. So it's a big, big business challenge for them.

I think we're out of time. Thank you very much!

Before it's here, it's on the Bloomberg Terminal.