Data reference model hits the streets
- By David Perera
- Nov 07, 2004
Data Reference Model
Agency enterprise architects can stop holding their breath. After months of delay and years of development, Office of Management and Budget officials released the document Oct. 20.
The first volume of the data reference model is widely considered the linchpin on which the entire federal architecture structure rests — and the most difficult part of the enterprise architecture. The model will be issued in four volumes.
The first volume sketches a basic data model based on three concepts: categorization, structure and exchange of information.
"It focuses on the needs of information exchange and does not try to set data storage structures and formats," said Kim Nelson, chief information officer for the Environmental Protection Agency and co-chairwoman of the CIO Council's Architecture and Infrastructure Committee.
Reactions from those who reviewed the document ranged from relief to muted frustration, but few doubt that the stakes are high.
"If you don't have a good grip on your data, then all the computing power in the world and all the fancy applications in the world don't solve your problems," said Mike Tiemann, a principal at AT&T Government Solutions and a former agency chief architect who helped create the federal enterprise architecture.
OMB officials chose to approach cross-agency information sharing cautiously. "They've simplified it considerably from even previous draft versions," Tiemann said. "You've got to walk before you can run."
Nelson said the first volume provides a high-level strategic framework.
"It is a first good step in that it starts the process," said Debra Stouffer, vice president of BAE Systems Information Technology and a former federal chief architect. "Would we like more detail? Absolutely, and I think it'll come."
Other experts expressed similar reactions to OMB's efforts. "I think the overall structure is good," said Michael Daconta, the Homeland Security Department's metadata program manager. However, he added, the model "does not follow modern principles of knowledge representation" and forces agencies into guesswork.
"Some of the language is too general to be implemented correctly and consistently," he said. Without revision to fill some of the model's gaps, he added, agency confusion may be inevitable.
Daconta said the data reference model still needs a security context. The first volume says, "Security requirements must be considered at each level," but it doesn't offer agencies a way to do that.
"That's a significant omission that will need to be corrected very soon," he said, adding that it can't wait for future volumes.
Some say the data model needs better-defined links with the other reference models in the federal enterprise architecture. "You've got to work in an integrated fashion," Tiemann said.
One general consensus is that OMB officials must take a leadership role if the architecture is to take hold. "This is not a three-year job. This is a 15- to 20-year job," said Charlie Grymes, the Interior Department's Recreation One-Stop project manager.
Architects acknowledge that the first volume needed to be written from a high-level perspective. "If you ever want to find a topic that will cause a manager's eyes to glaze over, this is it," Grymes said. But the missing and promised detail that OMB officials say will be revealed in future volumes is needed quickly, he said.
Some question whether an architecture can be created in such a fashion, with each reference volume adding yet another piece. "My one recommendation would be to look at those [volumes] and make a determination whether or not they could perhaps cut the four into three [to] get them out a little sooner," Tiemann said.
OMB officials and CIO Council members have made completing the remaining volumes a high priority, Nelson said.
"Certainly, they have recognized that there is more work to be done," Stouffer said.