News Stay informed about the latest enterprise technology news and product updates.

To avoid enterprise data mashup madness, plan ahead and keep it simple

Enterprise data mashups make integrating data quick and easy for business users. But without proper planning, mashups can be a nightmare for IT.

Enterprise data mashups may not have made their way to most organizations yet, but they're coming, and IT had better be prepared.

The enterprise mashup market will reach nearly $700 million within the next five years thanks in part to mashups' ease of use, according to Cambridge, Mass.-based Forrester Research. Mashups allow business users to integrate data from multiple sources – the Web, internal databases, Microsoft Excel spreadsheets – without having to write any custom code and without the assistance of IT.

An enterprise mashup tool might allow a commodities trader to integrate a company's sales data with the price of gold, then display it in real time in a widget on the desktop, for example. Or inventory data could be mashed up with corresponding warehouse space and presented on a map for easy analysis.

But despite their relative simplicity, or in some cases because of it, mashups can cause headaches for IT.

"The very thing that makes mashups so beneficial – that business users can create them on their own – will likely come back and bite [IT]," Michael Gualtieri, a senior analyst with Forrester, wrote in a recent report.

More on data integration
Find out what a "bashup" is and how it can improve your BI capabilities  

Get the real deal on data integration for business intelligence  

Find out why companies are choosing real-time data integration over batch-oriented techniques
The problem is that business users, lured by the ease of use of enterprise mashups platforms and applications from vendors like Chevy Chase, Md.-based JackBe Corp. and Redwood City, Calif.-based Serena Software Inc., are still likely to find themselves overwhelmed at some point -- confused by multiple data points, for example -- requiring IT to come to the rescue.

"They'll start building a mashup, and then there's this one final thing that they can't do, and then they'll have to call [IT], and that kind of blows the whole benefit right there," Gualtieri said. "The whole point is they could do it themselves and they wouldn't bother IT."

Inadvertently creating a monster application that becomes critical to the business is another likely scenario. "The [critical mashup application] probably wasn't built in a scalable fashion. There's probably a lot of problems with it, but it met some need that IT didn't meet," Gualtieri said. "Now, IT is going to have to take it over."

There are, however, proactive steps that IT can take to limit the chaos.

Training and simplicity key to avoiding mashup mayhem

To avoid these and other complications, IT should set realistic expectations of what mashups can and cannot do, Gualtieri said. Users should be provided with "in-context training," as he puts it, to understand what data sources and services are available to be mashed up.

"Users get seduced by how easy the tool is to use. But the training that they need is not actually how to use the tool, because that's easy -- it's been designed to be easy," Gualtieri said. "The training they need is on how to navigate all the services and data sources that are available in their enterprise."

More important, he said, is to teach users how data relate to one another. It's relatively simple to conceive of how real estate listings relate to a point on a map, for example, but understanding how a stock price relates to inventory data is more complicated.

"Corporate data sometimes has less obvious key fields," Gualtieri said, and users need to understand them in order to make the most of mashups. "Part of a mashup is discovering relationships that people didn't think of before."

One way to make relationships between data and data sources easier to understand is simply to limit the number of data sources business users have access to, Gualtieri said.

"In your service-oriented architecture [SOA], there might be a 'get customer' service that returns all the information on the customer and that might return 172 data elements," he said. "That could just be completely overwhelming for most users."

Instead, IT could create a virtual service on top of its SOA that limits those data elements to just the 40 most critical ones, for example. "So all of a sudden you've created a virtual service that returns 40 data elements," he said, "and it's much easier for a business user to understand those 40 elements. The other 132 may have been more IT-related anyway."

As mashups continue to make their way into the enterprise – and they will, thanks to user demand, Gualtieri said – he also advises IT to:

  • Add mashup-related issues to the help desk escalation process. Even though the point of a mashup is to avoid users relying on IT, IT still needs to be prepared if and when an issue arises.
  • Create a support structure to ensure that mashups, especially those critical to the business, are continuously up and running. "The last thing you want to happen is for business-critical mashups to go down," Gualtieri wrote. "Ensuring that the mashup platform is operational will avoid costly outages."

Is your company experimenting with enterprise mashups? Email editors with your story!

Dig Deeper on Data warehouse project management

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.