Without an effective way to manage data warehousing meta data, companies cannot automate the processes involved in running a data warehouse or give end users a convenient way to understand the origins and nature of the data in the warehouse.
Although many companies are aware of the importance of meta data, few are willing to invest the time, money and resources to implement a robust meta data management system. Part of the problem is that the benefits are difficult to quantify and that design issues always take precedence over maintenance issues of which meta data makes up a critical element when building a data warehouse.
However, the biggest obstacle is the lack of meta data standards. The absence of standards, in effect, forces each customer to become a systems integrator, meldi ng meta data stores from disparate products that comprise their data warehouses. In essence, this requires companies to build a second data warehouse a data warehouse about the data warehouse, if you will. Few companies have the skills, energy or time to build two data warehouses at the same time!
Meta data standards are critical for helping companies achieve the goal of "lights out" data warehousing. They will also play a critical role in the emerging area of corporate portals that promises to give users transparent access to related sets of information and to application resources inside and outside the company.
Currently, there are two meta data standards competing for preeminence in the data warehousing and business intelligence domains. One was developed by Microsoft and is now managed and updated by the Meta Data Coalition (MDC). Another is being established by IBM, Unisys, Hyperion, NCR, Oracle and a growing list of other firms under the auspices of the Object Management Group.
Both groups have made overtures to begin the process of aligning these competing standards. However, time is running out for a formal convergence of the standards, leaving the industry to contemplate a meta data schism that can only be spanned through less-than-adequate, vendor-specific meta-model bridges and interchanges.
The Open Information Model (OIM)
Microsoft's OIM specifies a meta-model for component modeling, knowledge management and application development as well as data warehousing, among other things. OIM consists of 200 object types and 100 relationships described in the unified modeling language. It uses SQL as a query language and calls for XML as an interchange format between OIM-compliant repositories such as the Microsoft Repository. Microsoft developed OIM with help from 20 software vendors. Microsoft currently ships OIM with SQL Server 7 and Visual Studio 6.0.
In an effort to convert OIM into a de facto industry st andard, Microsoft turned over the fledgling meta-model to the MDC in late 1998.
The MDC is a coalition of about 50 companies (mostly vendors) dedicated to providing an easy-to-deploy solution for accessing and interchanging corporate meta data. In 1997, the MDC released the meta data interchange specification (MDIS) 1.1 standard. MDIS specifies a lowest common denominator meta-model and uses an ASCII-based data interchange format.
Although it is relatively easy to support compared to other meta data models, MDIS 1.1 has not been widely deployed. A total of seven vendors have supported MDIS, including NCR, IBM and Evolutionary Technologies. But several of the initial MDIS supporters, such as One Meaning and LogicWorks, have been acquired, so their ongoing commitment to the standard is in question.
To reenergize its flagging meta data standards initiative, the MDC eagerly embraced the OIM as its next generation meta-model and interchange specification. Unlike MDIS, OIM would pr ovide programmatic access to a meta data repository, enabling development of lights- out management tasks, such as synchronizing meta data across tools and other change management functions.
The problem with OIM, however, is that Microsoft specified the use of COM APIs for accessing and manipulating meta data from OIM-compliant repositories such as the Microsoft Repository. Therefore, in an effort to convert OIM into a de facto industry standard, Microsoft stripped out all the COM dependencies from OIM prior to turning it over to the MDC this year. This makes OIM "vendor neutral and platform independent," says Steve Murchie, Microsoft's product manager for SQL Server marketing.
To complete the transformation of OIM from a proprietary to an open solution, the MDC needs to map the MDIS specification to OIM. This is a straightforward task, since OIM is a superset of the elements found in MDIS. In addition, MDC needs to adopt/revise Microsoft's XML Interchange Format, which was devel oped for use with the Microsoft Repository.
Besides this work, the MDC is currently reviewing three new models to extend OIM. These were put forth by Microsoft.
- The business engineering model provides a way to formally document the structure and processes of an organization.
- The business rule model provides a way to capture, classify and store business rules used in various applications and business processes.
- The knowledge description model provides a thesaurus or common vocabulary for describing information contained in a taxonomy of end-user resources.
Common Warehouse Meta Data
The other meta data standard is being spearheaded by IBM, Unisys, NCR, Hyperion Solutions and Oracle under the auspices of the Object Management Group (OMG). Other vendors have recently joined the initial working group, which is open to any OMG member. These are Genesis, UBS and Dimension EDI.
OMG's RFP. This working group came toge ther late last year after the OMG issued an RFP for a common warehousing meta data (CWM) standard that conforms with OMG's existing framework for defining meta-models. The RFP calls for a meta-model defined using OMG's meta object facility (MOF) and a meta-model interchange mechanism based on OMG's XML meta data interchange (XMI) specification.
The MOF is a subset of the unified modeling language (UML) that is used to precisely define meta data models (a "meta-meta-model," if you will). It also defines a set of IDL mappings to meta-models as well as a set of CORBA interfaces that can be used to define and manipulate MOF-compliant meta-models. The IDL interfaces automatically generate Java, COM, C, C++, Cobol, Smalltalk and Ada interfaces to MOF-compliant repositories. The interfaces can be used to automatically generate repository server implementations from MOF meta-models or support repository interoperability.
The working group is also developing a meta data interchange standa rd based on XMI, which was ratified by the OMG in 1998. XMI defines rules for coding and transferring MOF-compliant meta data into XML document type definitions (DTDs) and XML documents. The working group is defining a data warehousing-specific DTD for exchanging meta data between data warehousing tools and repositories with MOF-compliant meta- models. Since XMI is already adopted and supported in commercial products, the XML DTD for CWM will be automatically generated by a number of tools.
The due date for initial submissions to the RFP was September 15, 1999. Final submissions are due in March, 2000.
Converging the Standards
The CWM working group has made overtures to the MDC in an effort to ensure convergence between OIM and the proposed CWM standard. Led by Unisys, IBM and NCR, the CWM working group agreed to use OIM as one of the design reference points for the new CWM meta-model. The other design reference points are JDBC, SQL92 and SQL99 draft sp ecifications, and implementation of IBM, Unisys, Oracle, NCR and Hyperion database, data warehouse and repository technologies, according to Sridhar Iyengar, an OMG representative from Unisys.
To further foster convergence, the CWM working group has brokered a membership exchange between OMG and MDC. Consequently, OMG is now a member of MDC and vice versa. In addition, the CWM working group has encouraged MDC to submit the OIM in response to the OMG RFP. Such a submission would likely set up a process that converges the OIM and CWM. As standard practice, OMG typically requires groups with strong competing submissions to converge their proposals and resubmit a single specification to the OMG for approval.
"We want to avoid the specter of having diverging standards," said Iyengar, who was instrumental in brokering the union between MDC and OMG. "The OMG has successfully unified modeling standards by adopting UML and can do the same with data warehousing standards by fostering pragma tic collaboration between OMG and MDC."
Platinum Submission? This spring, Platinum technologies had indicated that it would submit OIM to the OMG. Platinum is a member of both the OMG and MDC, and it has been a key Microsoft ally in promoting the Microsoft Repository, which uses OIM as its underlying meta-model. Unfortunately, Computer Associates' recent acquisition of Platinum raises doubts about whether Platinum will follow through on its intentions. At press time, it was uncertain whether Platinum or any other OMG member would submit OIM. However, both Computer Associates and Microsoft are members of the OMG.
Microsoft says it, too, would like to converge the competing meta- models, but it can't support a standard that is platform dependent, as is the emerging CWM or its XMI-based interchange. "XMI is based on MOF, which specifies CORBA, and CORBA is not platform neutral; it's not a natively supported environment in a COM world," said Microsoft's Murchie.
However , other representatives of the MDC have seen demonstrations in which XMI does not require CORBA. Consequently, they suspect that MDC will endorse XMI as another viable XML interchange mechanism, along with the Microsoft-developed XML DTD. "Generating and parsing a few different XML DTDs will not impose a significant burden since there are a lot of tools available [for doing this]," said David Marshall, an MDC representative from ETI.
Richard Soley, president of OMG, emphasized that IDL interfaces for MOF make it natively supported within COM as well as many other languages. In addition, XMI is simply a mapping of OMG meta-models to XML DTDs and document streams, and it doesn't require CORBA.
"The bottom line is that OMG meta-models, such as UML, are already platform independent and vendor neutral and managed by an organization with a formal and open process for creating industry standards," said Soley.
Unisys's Iyengar added, "The use of the MOF model does not dictate use of CO RBA. Unisys has deployed MOF-based meta-models using COM without any CORBA middleware."
Although it would be ideal to converge OIM and CWM within the OMG standards- making process, all will not be lost if this doesn't happen. It's likely that vendors such as Unisys and others will develop OIM-CWM bridges to automate interchange between dissimilar meta-models.
However, time is running out for a formal convergence of the two potential standards. Initial submissions to the OMG were due by September 15. Without a competing submission from MDC, the industry will have a meta data schism that can only be spanned through less- than-adequate, vendor-specific meta-model bridges.
To avoid this scenario, an OMG member needed to submit the OIM by September 15 so that a joint working group can be formed to merge the two competing meta-models and interchanges. (Since press time was prior to September 15, I can't comment on whether or not such a submission occurred.) If this happened, it's a good bet that a converged specification can be submitted to the OMG by its final deadline for submissions in March 2000.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access