Getting Data Integration Done with SOA Standards Lowers Cost, Speeds Completion and Improves Reliability >

By Bill Miller
April 27th, 2007
David Linthicum is a smart guy who has been around data integration for a long time. The title of his recent post in his SOA World blog, “SOA and Data Integration - The marriage of data integration and SOA could end up in divorce” is a little puzzling. I agree with David’s point that:
“… many data integration solutions are more about information/data than about sharing services, so they’re hard fit for many SOAs.”
However, SOA standards and methods are core to better Data Integration. As Loraine Lawson points out in her ITBusinessEdge post, David agrees with this, which he says in “SOA could become the tool for operational and aggregated data”, even though his choice of title for his most recent post may lead you to believe otherwise. The argument, like much of the problem, may only be semantic. David is talking about older Data Integration methods. There are a few leaders who are innovating by using SOA standards at the core to implement next generation fast and flexible Data Integration.
Data Integration is all about enabling many different data systems with many different data structures, formats, naming conventions, and access methods to share and exchange data with each other or with many different applications. The need for Data Integration stems from two desirable realities, distributed IT and innovation and change. Data is too valuable to be left behind, so the old must connect with the new. Distributed data must have the ability to be re-shared, re-combined and re-invented. Data Integration can be and has been accomplished in many different ways through the years; a situation which in itself leads to even more integration, re-integration of past integration, and integration built on top of integration (that is a lot of integration!).
If not done right, and it often isn’t, this can get very confusing, brittle, and is very expensive. Some analysts estimate that between 40% and 70% of all IT project costs today are consumed by integration. What can be done to tame the data integration problem?
The ideal would have,
1) All data available in a standard common format which is self-describing with respect to its organization and meaning (e.g. “XML”)
2) All data accessible in a standard common way (e.g. “Web Services)
XML and Web Services are two of the group of standards that are core to Services Oriented Architecture (SOA).
You must be using these standards, not just on the edges, but at the core of the way in which you approach data integration. In fact, I would encourage people to use SOA standards at the core whether or not the purpose of a data integration project is to actually enable enterprise SOA. Most integration projects today do not yet have SOA as the objective, but for us it is the means to be better solution. Data Integration with SOA standards to normalize and expose data, lowers both current and future costs, provides a better documented greatly simplified and more reliable system to maintain, and paves the way for future IT-enabled business innovation.
Bill Miller, billm@xaware.com

XAware blog

I am now also posting a blog on XAware's web site (www.xaware.com). I will replicate those posts here as well.

Bill

They're back!

Storage start-up Isilon successfully completed an inital public offering this past week pricing the company at $800 million and then trading up on opening day to $1.4 billion. The company reported $40 million in revenues through three quarters of 2006. WAN appliance company Riverbed, RVBD, also with $40 million in revenues through three quarters of 2006, priced its IPO September 21 at $650 million and has traded up to $1.5 billion. After a six year tech IPO nuclear winter, and prognostications that we would not see the tech IPO market again in our lifetime, it is back with gusto. This could bode well for venture capital investment in tech start-ups if the IPO trend hold for awhile. Already the ISLN numbers seemed to have an impact on breaking at least one barrier in a tough valuation negotiation that I know of. I am told that ISLN and Riverbed marketed their offerings using agressive forward revenue forecasts. A miss from would put a quick end to this new tech IPO bull market. Let's hope they hit their numbers.

The Enterprise Software Business Is Dead - as we knew it...

It is getting more and more difficult to get enterprise customers to pay for software licenses, especially for anything new (ie. not already part of their "stack") or from a start-up. The open source movement, the movement toward building custom software using off-shore develoopment, and now SaaS have conditioned IT organizations to pay for services, but not for licenses. The enterprise software license model is dead, replaced by a services model. But services, usually a combination support subscriptions and implementation, integration, development services, will not scale as well as pure software license models. Most importantly, the incentive to invest in innovative software development will be diminshed. It is hard to justify paying for software R&D when you have to give the results away just to get services revenue. I, for one, am mourning the passing of the pure software license model.

Let's talk ....about SOA

SOA is a hot topic, so why isn't it a hot business. Everyone I talk with who is involved with companies, both start-ups and big guys, selling some kind of SOA tool set, or platform, or whatever, complains that they aren't selling much yet, and aren't really sure why. There are lots of theories about missing pieces, long planning cycles, and difficulty getting the buisness guys to understand the need to invest in it. Gartner puts SOA right in the middle of the "trough of disillusionment". That is probably right. SOA will be very good for enterprise, and extra-exterprise productivity and agility when it does move from hot topic to real adoption. It will allow appicaitons to be more flexible and keep costs of improving business applications down, helping to alleviate two really big drags on business productivity improvements.

If you are a tech-type and want to play with building web services that front end a composite of various applications and data sources, you can download a free development and test environment from XAware (one of my portfolio companies). Go to http://xaware.com/downloadregistration.aspx

Milton Friedman

I was at Stanford a few weeks ago. I dropped by the Hoover Institute to look for Milton Friedman. I thought I would just drop in to introduce myself and say hello. Dr. Friedman was not in. Too bad, I said, I'll stop in again when I next get to Stanford. I was saddened today to learn that I will never have that opportunity. I am greatly disappointed to have never met the great man. I know many who have. All tell of a man of such superior intelligence, thoughfulness, and comprehension that he may have been of another world. I guess I hoped that, in meeting him and shaking his hand, some of that might somehow rub off on me. Milton Friedman changed the world, explaining, in language all could understand, why government planning fails and freedom works. He will be missed.