Saturday, August 03, 2002

Last week Bill and Steve gave an interesting Microsoft strategy presentation at Redmond. I haven't worked there in years, and my focus may be myopic, but here's some thoughts.

"There were some competitors who dismissed this approach, the centricity of XML, the importance of Web services"I still don't get the importance of Web services. Or rather, I understand that but I don't understand the value-add of UDDI and WSDL.
"this is a software problem, one of the toughest software problems ever tackled, easily greater than tough engineering problems like getting to the moon or designing the 747"In some ways, I agree with this. If they're smart, however, the problem will be broken up into many smaller problems, which can be far more independent of each other than could be possible in a space mission or airplane. Software is very complex, but it can be extremely modular. Hey, XML can help with that!
"today think of your address book information that's in your buddy list, it's on different machines, it doesn't get up to date, somebody can change something, it doesn't automatically get to you, so the information management is not people centric, and .NET My Services, the direction there is to solve that"It's definitely not clear from this who "owns" your buddy list. That's the problem with making something centrally available. On the other hand, we've gotten over that anxiety with email - many people use hotmail for email without worrying about Microsoft owning their emails.
"It's important to think of the breakthroughs that were required once we got that connectivity really in two parts. One is how you represent the information and the other is how do you exchange that information. The first is about information formats and schemas and the second is about a rich set of protocols."Yes! It's not enough just to have the Internet, we have to define how data is represented, exchanged and used. There's much confusion on that score. Today somebody posted a proposal for a RPC protocol to the general IETF list, as if it were the protocol to end all protocols. Well, we can have layers and layers and layers of protocols, but there still needs to be an understanding about how the information is formatted and what it means to the sending and receiving computers.
"In a sense you could say it's the Holy Grail that computing has talked about for a long time, having applications written by people who don't have to meet each other, because the number of connections are too great for that, they don't have to trust each other, their system can run even if the other guy's system is unreliable or even if it's malicious you can make sure that the wrong things don't happen to that application."OK, this is totally bogus. .NET is not the holy grail of computing. It won't give computers the intelligence to talk to each other! Duh.
"Jet Blue is an example of a competitor in the airline industry who's decided they're going to keep their IT costs below 2 percent and that's against a fairly typical industry average there of about 5 percent."Wow. That's exciting! IT costs are too high. Cutting IT costs like that, if it's generally applicable, would rock for the world economy. Not that I believe the cost reduction is all Microsoft's doing, however.
Portals are mentioned, in the sense that a government (or I assume a company) can standardize on a portal and then plug all sorts of value services into it without planning a single architecture for all the services.That wish will not come true for 5-10 years. Since I've had recent experience working with and comparing portals, I can attest that they are so different there's a significant learning curve and even some architectural differences required to plug into them. Some portal products (Sybase's) are all about architecture, and locking a bunch of valuable services into one architecture.
"Federation, the idea that all these different systems can be connected together without their being any unique root, that is if two companies want to trust each other they can just federate."This is a cool thing, but not that new. We federate email, why not instant messaging? Competition amonst IM services has stifled federation in that area. I won't be surprised if IM stays in its isolated islands, or if more services are islands.
Sharing authentication information"What we announced recently is what we call "TrustBridge," where you can take that internal Active Directory and say OK, if I want to share these identities outside the company, how can I use this WS security protocol to, say, take the Intel Active Directory implementation and the Microsoft one and exchange that information? That's the corporate-to-corporate case..."OK, but does every TrustBridge require individual attention? It won't do me much good most of the time, because the IT departments of companies employing my collaborators couldn't be bothered setting up a TrustBridge with my employer.
"Notification is the idea that instead of going and finding information there are certain things you care enough about that you want to be notified."I've discussed this technology recently on this blog. It sure sounds from the way Steve talks about notification that they want to centralize it. E.g. rather than getting new-email notifications via IMAP, and appointment-reminder notifications via CAP, and document-unlocked events via WebDAV, they want to use one protocol to transport all these notifications, and one notification server to collect them for the user. That's not the way the standards are going, so MS is either going to drag the standards in that direction or be non-standard. Probably both to a large extent.
To remove barriers in communication media, to improve user centricity: "A big part of this we believe is bringing the voice world and the screen world together and you're seeing this in a number of real time things."I believe in this big time, but again, it's very hard. How does your IM client know what your email contact book is? These are so separate today, separate servers and protocols and clients, that it will be hard to unify.
"Outlook will evolve from being an e-mail client to being far more than an e-mail client."Yeah, you can see that happening -- its had calendaring and contacts and journal and tasks for years, recently a little buddy list functionality, next I imagine it will integrate NetMeeting
"SharePoint, you're going to see, is really a key part of Office."Ouch. SharePoint is really not standards based. In fact, open standards are really not a focus of this strategy. I guess this is the "extend" phase of "embrace and extend" the Internet. I hope the damage from MS not following open standards in key components won't be too great, but I'm pessimistic.
To break down barriers between islands of information on your own computer, e.g. so you can search across your files and email with one search command, they plan to unify storage of this information.This shouldn't be too hard, since modern file systems are half-way there, having dual streams so that the file contents can be stored with arbitrary amounts of metadata. But I'm not intimately familiar with the problems here.
"So you're going to see with the PC a lot more changes in these next three or four years than in the last three or four."That's really unlikely. Systems get larger and more unwieldy as time passes. Unless they've spent the last three years simplifying PCs rather than adding features (which they have not), the next three years are more likely to involve smaller and smaller incremental changes in PCs. Unless they're talking hardware form factors rather than software GUI, in which case a breakthrough may radically change things.

Summary: It's all about putting the data in XML to exchange via Web services (SOAP); that's .NET. SOAP is now built into Windows XP. Increased privacy, but at the same time increased authorization information exchange. Increased federation, but also increased centralization. "So there's a lot of direction here." No kidding!

No comments:

Blog Archive

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.