to LooselyCoupled.com homepage
 
 Weekly emails: how to advanced search
 
 Glossary lookup:

 

Loosely Coupled weblog


Friday, May 24, 2002

Web services spell death for packaged apps
According to a new report from analyst group Ovum, web services will kill off traditional packaged applications within five years. Neil Ward-Dutton, Ovum research director, said: "Traditionally, packaged software has been the only way to give users something that works. That is creaking now, and web services will be the nail in the coffin. Web services make integration easier," according to this story on Silicon.com.

Ovum is characteristically insightful in nailing the two distinct perspectives that cloud people's perceptions of what's really going on: "There is the technology perspective, which looks at it as a way of packaging software, and there is the business perspective, which sees it in the sense of software as a service or utility," Ovum senior analyst Christine Axton is quoted as saying.

It all reminds me of the first press release by ASPnews, which I penned way back in November 1998 (sadly no longer online): "An emerging model of online computing pioneered by application service providers (ASPs) is set to obsolete traditionally purchased software, argues a new report ... With the establishment of online rental computing, says the report, "[Users] will access the applications they need, on demand, from online providers who will charge them by the second for the precise value of the features and resources they choose to use." Plus ca change, plus c'est la meme chose.

posted by Phil Wainewright 3:54 PM (GMT) | comments | link
Web services networking options
Until shortcomings in security and reliability get written into web services standards, something extra is needed when connecting business critical applications together. That's where web services networking comes in.

Two startups are particularly well known in this emerging field. One is Grand Central, which I recently wrote up in an article just published by ASPnews. The other is Blue Titan, whose CTO was recently interviewed on WebServices.org.

These two companies are complementary, in that Blue Titan's product is for deployment inside a corporate firewall, whereas Grand Central's mission is to enable connections between separate organizations. The latter seems to me to be more in tune with the real opportunity of web services, but another competitor, Flamenco Networks — who I have not yet interviewed — seems at first glance to have a more decentralized and therefore potentially more forward-looking architecture.

At such an early stage in web services networking, no-one can now which approach will triumph. Other contenders are emerging from previous generations of content distribution networks (CDNs). Akamai and IBM recently launched a co-developed solution for application distribution, while Cable & Wireless is pitching to deliver web services networking under its 2 Way Web Services brand.

One other player that may be a surprising dark horse is the Novell/Accenture joint venture Volera. The company has been something of an also-ran in the content networking field, but its pedigree — as the brainchild of Simon Khalaf, an early pioneer in online services — suggests it may yet blossom as a web services infrastructure player. The company recently pledged "to deliver software solutions that accelerate, manage and secure the delivery of Web services." Its progress is worth watching. But while CDN practitioners are well versed in acceleration, it has to be said that, thus far, the more complex challenges of management and security have been more successfully addressed by the likes of Grand Central and Flamenco.
posted by Phil Wainewright 3:38 PM (GMT) | comments | link

Thursday, May 23, 2002

This is not a geek thing anymore
The transition to web services transcends technology, which means it's too important to leave it just to the geeks anymore. Last week's O'Reilly Emerging Technologies conference was a geekfest par excellence, and BEA's VP of engineering Adam Bosworth had plenty of geekery in his presentation of WebLogic Workshop, the vendor's new visual development platform for web services applications. But the most telling part of his presentation was earlier on in his session — and as far as I can tell, the geeks didn't get it. I wasn't there, but I got BEA to send me a copy of Bosworth's presentation, because there had seemed to be something missing from the accounts I'd read of it.

As I suspected, there had been something missing from the reporting — but it's there in the Powerpoint, on the slide where Bosworth describes the "totally different application model" being driven by the Internet's ability to support application-to-application integration, along with the emergence of portable computing, which Bosworth believes will be where the interesting early work will be done because there's less pre-existing infrastructure and more challenges that web services can solve.

Most of the accounts I've seen mention the three usual elements of this new application model that we expect from Bosworth:
  • It is based on self describing coarse grained messages

  • It is asynchronous

  • It is loosely coupled

But he went on to mention two more that no-one has bothered to pick up on. These are the really key ones, in my view:
  • It is biological

  • The heart of these systems is message processing, not database

I suppose the reason why this resonates for me with such urgent vibrancy is that it brings to mind something BPMI.org chair Howard Smith mentioned to me when we were discussing the Business Process Modelling Language the other week. He said that BPML is founded on pi calculus, which uses mathematical types used in the study of mobile systems. In other words it's based on the same maths that applies to systems with biological characteristics, rather than machine-based systems.

So this is my takeaway from putting these strands together: The transition to web services, and BPML, and autonomic computing, and all the other stuff that's going on at the moment, is all about moving technology out beyond machine-based terms of reference into the human sphere, and making it work to human rules, not machine-based rules.

Or it should be, unless the geeks stay in charge. According to Bosworth, ".NET is a web services framework which is tightly coupled, synchronous, and less than highly available." Microsoft's architects have framed it to conform to their machine-based rules, not to the loosely coupled biological framework that web services really need. Here are some other points from another of Bosworth's slides to mull over:

"Web Services isnít about B2B:
  • It is about App to App

  • It is to EAI what Sql was to hierarchical databases

  • It is to portable computing what databases are to client server

  • It will ultimately change even the UI we have to one that supports push (the browser is an application)

  • And, today, we are where Client Server computing was in 1992/1993"

In a pre-conference interview, Bosworth made this comment: "At this point, the only point our customers need walking through at all is the loose coupling. Thinking through, 'What do you do for a loosely coupled architecture, why is that so important?' It's something they're still working on." Now I have a clearer understanding of why there's so much for customers to get their heads around.
posted by Phil Wainewright 2:01 AM (GMT) | comments | link
Using the web without even looking
Email gives us a lot of clues about how the web will be used for applications, and the most important of these is that the web will become the infrastructure, rather than the destination. "Indeed, future generations will probably look back and wonder why on earth we spent so much of our time idly looking at the Web instead of actually using it." The quote is from my column this week on ASPnews, where I mention other emerging characteristics that email shares with web services applications.
posted by Phil Wainewright 1:18 AM (GMT) | comments | link

Tuesday, May 21, 2002

Analysts start to get to grips with BPM
There's been a sudden clutch of analyst reports advising clients on strategies for business process management (BPM), which suggests this concept has finally begun the slow journey into the mainstream.

The first sign that BPM is becoming trendy comes in the form of a report from Forrester Research, always a reliable arbiter of CIO cool. The Right Way To Manage Business Processes is a briefing paper that recommends CIOs "should organise a process savvy team to kick off their first BPM initiatives today." Infoconomy's always useful web services newsletter reports that "[Forrester] analyst Sharyn Leaver says that the time is now right for CIOs to plan the architecture for cross functional processes such as 'source-to-pay' or 'prospect-to-cash'," although Infoconomy notes that Forrester's advice remains on the fence when it comes to selecting the right tools for the job.

Aberdeen Group analyst Darcy Fowkes also recommends CIOs should start investigating BPM. Her briefing notes that in addition to BPM pureplays, many vendors that use the BPM label come form either a workflow or an EAI background. But their solutions often lack the real-time adaptability of pureplay solutions: "Support for this kind of flexibility is what differentiates best-of-breed BPM solutions from traditional business modeling techniques as well as application and data integration platforms."

Finally, BPMi.org chair and CSC executive Howard Smith points out two reports by AMR Research. One delves into some of the differences between workflow and BPM, presumably with some success, since Howard quotes some comments from the report on "necessary components" built into BPMi.org's Business Process Modeling Language (BPML). The second is an analysis of CSC's e3 architecture for building a process-based view of business automation infrastructure.
posted by Phil Wainewright 1:14 AM (GMT) | comments | link

Monday, May 20, 2002

Novell set to squander its way into web services
According to this CRN article, Novell is ready to spend up to $700m to bolster its web services portfolio. The company has "better tools, a policy engine, app server" on its shopping list, according to the report — although it then goes on to state that "BEA Systems, Iona, Tibco, Vitria, webMethods and Mercator are among Novell's potential acquisition targets," only the first two of which would come anywhere close to fulfilling its claimed wish list.

The charitable interpretation of this story would be that it is the reporter who lacks a full grasp of the web services arena. However a far more likely explanation is that Novell's leadership has not the faintest clue what it is talking about and is merely grasping at a trendy web services label in the vain belief that this will restore its fast-fading fortunes. All of this is a great pity, since Novell really does have some very strong web services infrastructure technology, in particular its eDirectory platform. But buying up one of the EAI vendors is not the way to gain web services expertise, any more than buying up a professional services company has anything to do with positioning for the next wave of service-led computing. Unfortunately, having already made the latter mistake, Novell is almost certainly now destined to compound its error by committing the former as well.
posted by Phil Wainewright 3:18 PM (GMT) | comments | link

Assembling on-demand services to automate business, commerce, and the sharing of knowledge

read an RSS feed from this weblog

current


archives

latest stories RSS source


Headline news RSS source


 
 


Copyright © 2002-2005, Procullux Media Ltd. All Rights Reserved.