Implementing SOA without Enterprise Mashups? You might as well kiss your job goodbye.
Ok, ok, its an overstatement. But the ROI of SOA is difficult, at best, to define and measure. Have you noticed that the press and blogosphere is filled with SOA implementers/analysts discussing the ROI of SOA and the idea that stand-alone SOA efforts are DOA? For a small snapshot of this teacup tempest, look no further than the recent commentary from SOA expert David Linthicum, the Nucleus Report on SOA ROI, and the subsequent commentary from ZDNet's Joe McKendrick and IT advisors Neil Macehiter and Neil Ward-Dutton.
While these experts differ on issues like the importance of SOA ROI, how to calculate SOA ROI (if at all), and why we don't have more/better of it, they all seem to agree on one thing: 'Enterprise-wide support for SOA hinges on the ability to demonstrate value to the business at large — more growth, revenue opportunities, and all that good stuff.' (Joe's words, not mine.) And that's where your job is at stake. Or, at least, the long-term support of your SOA efforts.
Let's face it, SOA is plumbing. Nice, shiny, efficient plumbing to be sure, but still plumbing. And your average business dude/dudette (think sales manager, marketing director, finance officer, or customer support rep) could care less about it. In fact, if they think about it all, they probably just hope it stays right where it is: out of sight and running quietly. These same business folk probably appreciate the marble floors, wood-paneled doors, and brass fixtures that surround this plumbing much more. In other words, they like that bit of 'stuff' that actually frames the plumbing and brings it to life.
SOAs need to change this inward-focused quality. To paraphrase Macehiter/Ward-Dutton in their recent note, 'More big vs small thinking: SOA vs BPM', IT must focus on where the real business value of SOA lies. That means it needs help. Macehiter/Ward-Dutton point out that BPM can help distill some SOA value up to the business level. And as one of the early implementers of Enterprise Ajax, JackBe knows from extensive first-hand experience that Ajax makes a great service consumer.
To this collection of SOA-complimenting tools I'd add the enterprise mashup. JackBe has found in its enterprise mashup implementations that they can actually drag the SOA out of the proverbial IT basement and onto the end-user's desks. It's not only highly visible, but it's user-driven, giving IT a way fulfill the promise of SOA and enhance that elusive SOA ROI.
As a practical guy I like definitive examples. Luckily, we've already seen a number of very real synergies from the mashup/SOA combination. Just a few of them include:
- Mashups can help create normalized 'virtual' services from sources that haven't been 'SOAed' yet. It's no secret that SOA efforts can take years. Until the formal SOA magic has been applied, a quick, standardized service can help users get started earlier than otherwise.
- Mashups let users 'right-size' the granularity of services. Now IT doesn't have to guess/study/analyze whether a service offers data that is 'too specific', 'too general', 'too dated', or 'too cold'.
- Mashups let users share their resulting services, making them a part of the service-generating network. Now IT doesn't have to do it alone.
- Mashups let end-users visualize the SOA in graphs, charts, tables, maps, etc. Instead of hoping the aging corporate portal has a place/way to get services visualized in the way(s) the users want , users can each do it themselves to meet their own ever-changing needs.
- Mashups let users join in data from outside the enterprise. Today's SOA efforts are largely inwardly-focused. But users often want to include external data in their work. Mashups don't care and good mashup software makes the actual location of a data service irrelevant.
---
Parting Comment 1: Kudos to Geek and Poke for distilling this talk of SOA ROI down to a single cartoon.
Parting Comment 2: Coincidentally, JackBe's blog has talked in the recent past about the cost-of-ownership for enterprise mashups. We didn't do a survey or try to define formulas. But our multi-part blog on 'Enterprise Mashups and Total Cost of Ownership' (here's part 1 and part 2) is a good look at some of the business and IT impacts involved in enterprise mashups. (That's a biased opinion, I admit.)
4 comments:
One thing we have found also is that SOA is starting to get a skeptical look from the companies that are actually trying it. At the same time many of them are keeping earlier projects in place, some of which will eventually get on the SOA backbone, and maybe some of them won't. Using a Content Based Integration strategy (a defined mashup architecture with some governance and patterns thrown in) can actually leap frog most SOA projects in terms of delivering a capability that the users can actually recognize as a direct benefit.
A well thought out CBI framework can deliver information to the user in around 8-12 weeks, and most of the SOA projects we have seen or been a part of are 18-24 month death marches. What's more if the CBI approach uses a well structured REST interface as a standard you can also make them SOA endpoints, so when the Big Science guys finally catch up the information from the systems you have pugged in are actually waiting for them to ask for it.
Look at Corizon for real enterprise mash ups.
www.corizon.com
From what I understand of Corizon (and this may be old info), it's a very programmatic implentation of legacy apps. Jackbe is a phenominal visual environment that deploys auto-generated AJAX applications to the desktop. It is not only very quick to build new portals and mash-ups but very easy to maintain, update and deploy.
Corizon differentiates itself by service enabling the UI. This adds the structure and control required for deployment inside the enterprise, while creating an environment in which the UI compositions, or mashup, can be built quickly and safely.
For more, read this article: http://blog.corizon.com/2008/06/lightweight-or-enterprise-grade.html
Edwin van der Sanden
blog.corizon.com
Post a Comment