It's not difficult to do, but backing up BI data takes time and effort. Without it, that hard-earned data could float away.
Mechanically, the backup of business intelligence data is easy. More arduous is the crafting of cohesive plans to protect strategic, analytical information that drives vital business decisions.
There's good news for executives worried about the preservation of BI data. Backup utilities are bundled into most BI products, and use of these functions seldom requires help from vendors or consultants.
But there's some not-so-good news as well: BI backup isn't automatic. Instead, corporate officials must devote energy to key data collection and retention decisions that will affect backup strategies mightily.
And while most companies that have embarked on BI projects perform at least some backup, the largest companies are doing less BI backup because their data warehouses are so big, according to Gartner analyst Donald Feinberg. For instance, Wal-Mart Stores reportedly backs up almost no BI data because of the size of its warehouses, which contain terabytes of data.
"The trouble with the backup of BI data is that you are forced to think about it, whereas with transaction data, you aren't," notes Mayur Raichura, director of information systems at Long & Foster Real Estate in Fairfax, Va.
Raichura and his staff are now sharpening Long & Foster's BI data collection policies -- moving away from initial attempts to capture metadata about every search conducted against the company's site. If all that data were to be captured, Long & Foster would generate more than 30 million rows of data in a single year, says Raichura. "How fast you decide to delete -- or not to delete -- your BI data and how you collect this information will impact your backup strategy," he says.
The first issue to think through is whether particular BI data sets need backing up at all. Consider the company that loses strategic BI data that's factored only into high-level decisions made at quarterly or annual meetings. It might be fairly easy to reproduce that lost data, because the information is highly aggregated and not very detailed, according to Keith Gile, an analyst at Forrester Research Inc.
On the other hand, tactical and operational data feeds more immediate decisions and likely requires more-hearty backup measures. "Tactical decisions are made over days or weeks and are more associated with specific business cycles. Under these circumstances, BI data backup is necessary," says Gile.
Having decided where to focus backup efforts, BI executives should then begin shaping BI data maintenance policies into sound backup strategies. It's an effort that can prove challenging for even the most seasoned organizations.
"We are pretty mature in this area," says Grant Felsing, decision support manager at Briggs & Stratton, a Milwaukee-based engine manufacturer. "In a lot of the conversations I've had with other companies, I've come across a lot of blank looks."
Briggs & Stratton's data warehousing infrastructure uses Unix-based servers running software from Cary, N.C.-based SAS Institute to churn BI data from dozens of operational sources into executive management information. Briggs & Stratton gathers transactional data from financial documents and backs that information up on a monthly basis.
Backing up this BI data is technically simple but is no rote exercise. "While the Unix backup tools are pretty thought-free, you've got to decide on the timing of the backups and set them up based on standard models that extract data from various sources," says Felsing.
For companies with less BI savvy, there's also the risk that BI backup routines will be neglected. "My feeling is that not too many people are going through these exercises because they are so new to warehousing. But a company is smart not to overlook this," observes Jody Porrazzo, director of econometric risk strategy at Apex Management Group Inc., a consulting and insurance services firm in Princeton, N.J. Apex collects metadata and uses it extensively in reporting operations.
Apex is deploying a new SAS BI server, and backup functions are included as macros in the system. "It's very straightforward and very simple," says Porrazzo.