Don't Let Data Quality Concerns Stop a Data Migration Project

Partner with business decision makers to ensure data quality during the data migration process.

Data migration projects may seem straightforward, but they often hit snags for a variety of reasons-not the least of which is data quality.

Bad data can break new applications or systems, according to experts, and in order to avoid pitfalls associated with it, it's necessary to loop in business managers to help IT staffers understand and prioritize the cleansing of mission-critical data.

"What they need to do is analyze the legacy sources early in the migration effort-measure the levels of quality and identify the quality flaws that will cause the new system to experience issues," said Ted Friedman, an analyst with Gartner. "Then make the decision of whether to clean up the issues at the legacy sources or while data is being migrated from old to new."

This requires heavy involvement by the business and not just its IT department, Friedman continued, explaining that the business executives know what qualifies as good enough in terms of data quality.

Click here to read about IBM's upcoming "Cheetah 2" data server with enhanced availability and failover recovery features.

Studies have found delays to be typical for data migration projects. A study by research company Bloor released in September 2007 found some 64 percent of all data migration projects come in late and 37 percent come in over budget.

"No company will pay for perfection and you can't get it anyway-for example, it is estimated that customer data deteriorates at around 1.5 percent per month [due to] name changes through marriage, moving house, getting a new cell phone and so on," said Philip Howard, an analyst with Bloor.

To Tony Sceales, chief technology officer of Celona Technologies, a successful migration strategy means incorporating a range of cleansing strategies at different points-some premigration, some during and some after. In an interview with eWEEK, he said data quality should be part of an overall data management strategy.

"What's important is to transfer ownership of the data quality conversation to the business," Sceales said.

Some of the more prominent data quality software vendors include Informatica, IBM, DataFlux and Business Objects. Stef Damianakis, CEO of Netrics, which specializes in data matching to catch inconsistencies, said he recalled his company working on a project with a customer in which only the IT department-and not the business decision makers-were involved.

"When we identified problems with the data, IT's response was, 'Can you come and tell the business people for us?' If the IT people and the business people aren't working together, the project won't go anywhere. Both skill sets are needed for a migration to be successful," Damianakis said.