Some believe that having too many different technologies is simply unmanageable for the average enterprise. “The problem is exacerbated with technologies that need a high degree of integration or customization with other products,” says David Gorbet, vice president of engineering for MarkLogic. “For example, a database containing unstructured text that doesn’t have built-in search capabilities will need to be stitched together with a search technology. As organizations attempt to stitch together multiple different technologies in different and nonstandard ways, it becomes harder and more costly to manage the infrastructure.”
In addition, Gorbet continues, data architecture and governance are inhibited in such environments. “It becomes much harder to apply appropriate governance to the data as pieces of data exist in multiple different systems with differing—and sometimes quite weak—governance models,” he says. “Just because there is a new class of data problems doesn’t mean that enterprises can afford to throw out 30-plus years of governance best practice. The technologies they choose to run in their data centers still require key features like security, manageability, backup and restore, high-availability, disaster recovery, and transactional consistency.”
The Fifth V of Big Data—Visibility
Imhoff doesn’t see issues with variety and innovation now being seen within the enterprise data center. Rather, there’s the matter of visibility of the data, and how IT and data managers can be more aware of where information is being sourced and applied. Business people have been going around IT to use new forms of databases, often from the cloud, but there’s nothing new with that trend, she points out. Rather, the challenge is being able to gain visibility into how the data is being used. “IT has never been able to control the information assets the business community uses,” says Imhoff. “They have their own databases, they have their own credit cards, they can buy cloud stuff, they can subscribe, they can do anything they want. And IT is completely blind to it—and therein lies the problem. If there’s a breach in security, and a compliance report gets published in The Wall Street Journal, who gets blamed? It’s not the business community, it’s IT.”
Large players are extending their offerings to address the challenges of managing unstructured data.
What is needed is greater visibility into enterprises’ information assets. “IT can’t control the assets, but please at least give them the ability to monitor them,” Imhoff says. “Who is using what data when, where and for what purpose? For example, IT needs to be able to see if a report being created is using data that is unsanctioned and ungoverned. They need to see if users are following security, privacy and compliance policies, and warn them if they aren’t.”
Management Blinders
When it comes to data and information issues, management does not understand that the challenge is coming, and fails to recognize the significance of unstructured data assets to the business. Enterprises need to determine up-front what the goals are of their data management efforts. Are they seeking to compete on analytics more effectively? Enterprises need to establish what goals they are attempting to achieve as they enter this space. Managers need to understand how to measure the benefits of new database deployments, as well as understand the potential return on investment from being able to leverage new data sources. Many of the new breed of data environments on the market can be accessed via the cloud, allowing for experimentation and research. In addition, skill sets are another challenge for organizations.
In the process, it’s essential to keep an eye on the business goals of the data. Ultimately, this may require an approach that employs both traditional relational and next-generation data environments. “Before being blinded by the technology, be sure to understand your business requirements,” says Steve Horn, senior consulting product manager for Dell’s Cloud and Big Data Solutions Group. Often, the business doesn’t need a lot of new technology fixes. “If your window for new analysis is measured in days, you may not need the lightning speeds from a product that provides microsecond response. If you need safe and secure transactions then that traditional relational database is likely the answer. The point is to make your database data store decisions based on business needs and fit the products to those needs. No one product can meet every need, so you will likely have more than one in your shop.”
Business First
Evaldo Horn de Oliveira, business development director for FairCom, outlines the way enterprises can go about determining what types of databases are best fits for the business problems at hand. This process starts with first ascertaining business needs and the type and volume of data they must store and use. “Assess whether there are other data sources, such as data in legacy applications—such as COBOL or other aging languages—which are not readily accessible, yet provide vital context for business decisions, before determining how to proceed,” he says. “It’s also important to assess how the database frameworks will be managed and adapted as business needs evolve and need to be maintained. As they pull these elements together for their enterprise, managers need to focus on ensuring all the elements can be connected and managed through standard APIs.”