Turning data into useful information requires knowing where it is, being able to access and analyze it, and of course, having one version of the truth. Cloud data services can help by providing self-service access to integrated and mashable information from siloed information systems.
“A data cloud is about being able to find and access data in its native location; not moving a copy of data into some warehouse,” said Lyn Robison, an analyst with Burton Group, a division of Gartner Inc. “Ideally, a data cloud will have some analysis capability from a BI standpoint. It’s a different approach—instead of bringing data together, it’s a concept that says keep data in place and make it accessible.”
Formally, Burton defines cloud data services as “patch panels for data that connect individual data items between silos.” … They provide “a means of reconciling, virtualizing and integrating computer-based information that resides in silos, and making that integrated information appropriately available online for search, query and analysis by businesspeople in a freeform manner using Web browsers, mobile devices, off-the-shelf software applications and analysis tools.”
Key to a data cloud is what Robison calls “entity and identity resolution.” A common problem with data warehouses is “you bring customer information into a data warehouse from different information systems; the catch is each customer is IDed using a different identifier,” he said. “With a data cloud you still have the problem, but it solves it by recording the identifiers in a central registry and uses that registry for dynamic queries against existing source systems.”
Robison points to Data.gov and SKUforce.com as examples of data clouds. “SKUforce is a specialized data cloud that will hold data from multiple enterprises working together.” For example, he continued, “you have part numbers for products, bills and materials, subassemblies, and you have a registry on the Web that reconciles and serves as registry for each of those part numbers, and it can be used consistently up and down the supply chain. You can have external data clouds used between partner enterprises or internal within the enterprise.”
Robison said cloud data services focus on what he calls “fungibility” or making a distinction between fungible data (assets that are interchangeable) and nonfungible data (assets that are not interchangeable). “Our information systems have tended to treat everything as fungible. They haven’t kept track of the identifiers for these important assets. You can use a data cloud to reconcile the identifiers of nonfungible assets.”
For organizations that want to get started with cloud data services, Robison offers these tips:
- The philosophy of a data cloud is frugality. “You don’t have to buy software to get started or install stuff; you can start with the software licenses the enterprise already owns,” he said.
- Build a simple list of relevant business topics and which information systems hold information on those topics. Publish that on the enterprise’s intranet. “One client did that and that page quickly became the most popular page on the intranet,” Robison said.
- Record the usage of that page. “Gather both statistical information about how useful that page is and anecdotal stories,” he said.
- Look for a business problem caused by businesspeople being unable to find information they need on particular topic.
- Start to build a registry that gives more detailed information on that intranet page; “say, maybe a master list of airplanes contained in those systems. You start to give businesspeople an idea of what system do I look in, and what do I call the airplane in each of the various systems,” Robison said. Document the results anecdotally and statistically.
- As an organization gets more advanced, it can start to justify some investment. “Go slowly into things that require investment,” he advised.