IT Strategy Letter, June 24, 2002

Microsoft v. Java and Détente. What vectors have converged to make web services the hot new API technology? Here are my thoughts on one such vector: Microsoft supports web services, in part, because they give the company a way to bypass Java and provide services to the mobile-device market. [This essay is an excerpt from the first-draft manuscript of my forthcoming book. Posted here for comment, so please feel free to tell me what you think.]
Posted Thursday, June 20, 2002 7:44:30 PM

Reports from the O'Reilly Conference. Thanks to Brent Sleeper of The Stencil Group for posting links and comments to two articles that came out of the O'Reilly Emerging Technology Conference in May.

Robert McMillian reports on a presentation by BEA's Adam Bosworth. "Asynchrony, security, and reliability--these are the principles that will make or break web services in the next couple of years."

John Udell, as usual, provides a valuable state-of-the-technology report, raising issues including (a) the doc/literal versus RPC problem, (b) SOAP routing, and (c) security. My favorite quote: "Discipline is a scarce commodity." It's in the context of programmers' natural tendency to use tight coupling in conjunction with the RPC model.
Posted Friday, June 14, 2002 2:28:22 PM

Loose Coupling is Like Pornography. As someone once said regarding the latter, "I can't define it for you, but I know it when I see it." Some interesting attempts this past week to define and explain loose coupling from John Udell, Patrick Logan, and Phil Wainewright.
Posted Thursday, June 20, 2002 1:58:18 PM

Who Are the Players? What does it take to be a top-tier player in the web-services game? I think you need three things: (1) software products, (2) a sales/distribution channel, and (3) strong professional services.

IBM and Microsoft are obvious. As a company that derives half its revenues from products and the other half from services, IBM has all three bases covered. Microsoft derives only 5% of its revenues from professional services, but it makes up for that with its huge army of trained developers and resellers. Who else?

Remember those red-box folks from Utah? Novell isn't a name we hear as much about as we used to, and certainly not as one of the top-tier players in web services, but that could change. Desperate to recover its lost luster, Novell has picked web services as its new focus. On June 10, the company announced the acquisition of Silverstream for $212 million to enhance Novell's product line. Last July, the company acquired Cambridge Technology Partners, which gave it the professional services piece. It also brought on CTP's Jack Messman to replace Eric Schmidt as President and CEO. And what about the channel? Novell still has its own army of loyal--but desperate for a new vision--VARs. Yes, Novell may be a longshot, but one worth watching. [eWeek Interview with Jack Messman]
Posted Friday, June 21, 2002 5:15:00 AM

And more on Novell...

The Identity Race. Phil Wainewright writes, "Microsoft has not handled its entry into the digital identity sphere particularly well, and its rivals seem to have sensed that its vulnerability has left open an opportunity for someone else to seize the initiative." In his ASP News column last week, Phil suggests that AOL, Novell and Critical Path may be "poised to step in."
Posted Friday, June 21, 2002 4:59:55 AM

Another Good Executive Intro. This one comes from Michael Jarosik at Start Magazine and is posted on Grand Central Communications' site. A few highlights:

Jarosik explains that web services aren't the application but the framework.
He points out that for tactical implementations, you're likely to turn to existing vendors for your tools: your EAI vendor to add web services to an EAI package, or your application server vendor to add web-services interfaces there. However, once you're ready for more strategic deployment of web services, the investment requirements become much more complicated, and you may want to look elsewhere.
Mike Gilpin of Giga is quoted regarding the advantage of using a single technology for both application development and integration. [It's been my experience that these are often handled by separate teams using different tools.]
Jarosik points out that there are three ways for users to interface with web services: (1) via portals, (2) through legacy apps such as CRM, and (3) via alerts.

[Source: Brent Sleeper. Read his comments]
Posted Sunday, June 23, 2002 12:47:06 AM

Are Web Services Too Slow? HSN, Inc. (formerly known as Home Shopping Network) opted out of turning the search engine (200,000 searches per day) on its retail-oriented web site--hsn.com--into a web service, which was their original plan. Apparently they thought an architecture based on web services would be too slow for their application. But the article doesn't suggest that HSN actually ran tests or at what level in the architecture they considered using web services.

Also covered in the article is a case study of how American Electric Power in Columbus, Ohio, will be using web services to "pull data out of its PeopleSoft human resources system." This strikes me as merely using web services as a data-extraction API and not taking full advantage of a service-oriented architecture. But perhaps that's all you can do with packages like PeopleSoft for now, until they're re-written using a true service model.
Posted Sunday, June 23, 2002 1:13:53 AM

Tip of the Iceberg. This article (PDF) explains why SOAP, WSDL and UDDI only solve a small part of the problem, and that there's a a lot more work to be done. [Source: eAI Journal]
Posted Sunday, June 16, 2002 9:47:27 AM

The Basex Web Services 2002 Survey. "The results of this survey, targeting CIOs and others involved in IT strategy decisionmaking, will be used by Basex in an upcoming report, entitled "Web Services: Not Just for the Fortune 100". All participants who so desire will receive a copy of the survey results and the executive summary of the full report." [Source: Allen Weinberg, Glenbrook Partners]
Posted Friday, June 14, 2002 2:14:09 PM

Theory and Practice of Internet SLAs. Two weeks ago I had the pleasure of sharing a podium with Chris Overton, Keynote's guru of statistics. Here's a PDF of Chris' segment of the presentation. Two out of 13 pages are a bit geeky (cosines and differentials), but there's a lot of great food for thought. My favorite (and the source of a big Aha! for me) is the section entitled, The large cumulative effect of nagging little penalties. (And what a great title, eh?) This is where Chris explains why it's so important to have multiple SLA penalty thresholds rather than one big one. As he points out, if the vendor has already missed the threshold, what's to motivate it to keep trying? And if the vendor is likely to meet the objective, it also has no motivation to do better. More good stuff here.
Posted Sunday, June 23, 2002 3:09:54 PM

Major Changes at Loudcloud. I don't typically report news (others such as The WHIR do a better job of it), but this is a particularly noteworthy story. The highest-visibility MSP, Loudcloud, is selling its MSP business to EDS for $63.5M, changing its name to Opsware and focusing on its site management software of the same name. EDS is also taking a license for Opsware for a whopping $52M over three years. At the end of the day, will the pre-web outsourcers be the only ones left offering managed e-commerce services?
Posted Monday, June 17, 2002 9:02:57 AM

The next day, Intel announced it was getting out of the web-hosting business. Not selling it, just shutting it down cold-turkey.

The Economics of Peering. Rich Miller of Carrierhotels.com explains peering at The Web Host Industry Review.
Posted Friday, June 14, 2002 1:46:43 PM

Nothing planned for the rest of the summer. I've got my head down and my fingers glued to the keyboard, writing my next book, Web Services: Strategies for the Real World. Look for it at the end of the year.