Modeling Data ManagementExperts look back at the SEI Capability Maturity Model to find modern answers to modeling information and data maturityby Joe Celko A lot of good data management ideas floated around the 14th Annual DAMA International Symposium at the end of April 2002, but one idea in particular stuck with me. You may remember the Software Engineering Institute (SEI) Capability Maturity Model. Academics at Carnegie-Mellon University came up with five phases that a software organization goes through as it matures and then rated organizations on this scale. The phases are: 1) Initial: You build software like you never have done it before and will never do it again. One hero spits out code and you don't worry about maintaining or documenting it. Whatever the programmer gives you is good enough for the end users. 2) Repeatable: You actually have a project plan, and the plan might even include some quality assurance, documentation, and things like that. 3) Defined: You follow the plan, which is at the organizational level rather than the project level. You expect to train people, have compatible software, and follow organizational standards. Think of skilled craftsmen following a blueprint and using the standards of their trade. 4) Managed: The organization follows the plan and measures the progress as it goes, similar to an assembly line for software. Managers know what's happening as it happens and the software is also monitored. 5) Optimizing: The final phase is when the factory becomes self-aware. The lessons learned on the project are used to prevent defects before they occur and manage technological changes. There's a constant organized feedback mechanism to improve the cycle time and product quality. Unfortunately, very few organizations have gotten past the second stage and some are even slipping back to the first stage. These days, this devolution is called "extreme programming," "fast cycle time," or some other buzzword. Some people actually say it's a good thing. How soon they forget history. Axes of QualitySeveral speakers at the DAMA symposium spoke about building a maturity model similar to the SEI model for data and metadata. This topic was the subject of a presentation by Jeremey Janzen, senior data administrator at the British Columbia Ministry of Forests. Janzen had a two-axis model based on process and information maturity. He took the five levels from the SEI model for the process axis, then classified information maturity as: 1) Unstructured: Everyone will do their own thing and not worry about anyone else. 2) Uncontrolled: Everyone will make the data fit if they have to, but nobody's sure where it is or what it means. 3) Shareable: The data is defined, but not necessarily immediately usable by another process. Organizations need a conversion process between them. 4) Integrateable: One process can use data from another process immediately because the data and the metadata are well-defined at the corporate level. From these axes you define data by moving diagonally up the grid. 0) Personal data: Includes daytimers, personal calendars, and browser favorites. This information is generated by ad hoc processes and is unstructured. 1) Local data: This data exists at the local office and nobody does much to negotiate common definitions across the whole organization. This information might be derived data used for a local project. 2) Extended data: There's a definition and some kind of database for it, but it's used at the local level and shared by going through the database. 3) Full corporate data: This data is defined, used at an organizational level, and has a custodian who's responsible for the quality of the data. This data can be specific to one application or shared among many applications. The advantage of these categories is that you can now think about where data should be. Personal data won't go away, but what if I'm hoarding information in my personal data that needs to be moved up the corporate ladder? Sounds like a level four or five process is needed to make the move possible. Joe Celko [71062.1056@compuserve.com] is an independent consultant in Austin, Texas and the author of Joe Celko's SQL for Smarties: Advanced SQL Programming (Morgan Kaufmann Publishers, 1999).
|
 |
Most Popular This Week
IE Weekly Newsletter
Subscribe to the newsletter
| |
|