In the past, there's been some buzz about how enterprise-class mashups could be used for data integration. This week, ZapThink's Ronald Schmelzer offered guidelines on how IT can shift from talking about mashups to actually supporting them.
He also made this important point: Mashups are not a replacement for or just another form of data integration:
For certain, the act of composition achieves the goal of integration, and the consumer-centric mashup is just an aspect of composition. However, data mashups intend to solve different problems than traditional data integration approaches, and in many ways, do not replace the need for or value of traditional data integration solutions."
What mashups can do is give employees a way to pull data together - integrate it, if you will - on the fly, which translates into fewer one-time reports requiring data integration for you. That may be a bigger deal than you'd at first think. According to Schmelzer, 80 percent of the value businesses derive from data comes from the 20 percent of "fixed, highly optimized data-integration approaches implemented over decades."
Likewise, he continued, 80 percent of data integration requirements are related to what Schmelzer called "situational data" - ad hoc requests that, while they may offer value, also cause 80 percent of your data-integration headaches.
So, maybe IT shouldn't think of mashups as data integration per se, so much as a tool for avoiding those annoying, spur-of-the-moment requests that require IT to run yet another report requiring data integration. And this is perhaps the simplest, albeit selfish, reason IT divisions should take a serious look at supporting enterprise mashups. Share the Data, Save the Ibuprofen In other words, if you share the data, you'll save the Ibuprofen. Or, at least, you'll offload the headaches to non-IT staff, and, really, isn't that kind of a karmic payback for demanding a Web 2.0 work world?
To do this, though, you're going to have add a data-service layer. In case you don't know, ZapThink is a SOA consultant group, so this is where SOA comes into play, according to Schmelzer. Essentially, you're creating data services that can be loosely coupled, rather than the traditional, tightly coupling of enterprise data integration. You're bringing the SOA style to your data, so users can mash it up.
In March, Carl Weinschenk shared a prediction that "...the emergence of a generation of computer-savvy employees is increasing pressure on IT to produce complex applications, including mashups, that are to their liking." Weinschenk pointed out this means IT will need to change to accommodate these expectations.
Schmelzer offered very specific guidelines on what IT needs to change. His recommendations:
First, the component data sources that can be composed must first be exposed as Service assets in a governed, managed, SOA environment. Second, the IT organization must give Service consumers the tools and methods they need to be able to successfully compose those Services with low cost and risk. Finally, the IT organization needs to shift itself from having sole responsibility for data integration to simply being providers of Data Services to be mashed.
"Change is inevitable. In a progressive country, change is constant," said British politician Benjamin Disraeli. Not everyone wants to be progressive, but often in technology, "progressive" often quickly changes into "expected."
Enterprise mashups are still very much cutting-edge, but given the value proposition, such as fewer headaches for you and more flexibility for business, you might want to contemplate embracing that change sooner rather than later.