Toward Universal Information Models in Enterprise Management

• Authors: Jean-Philippe Martin-Flatin
• Venue: DBTel '01 Proceedings of the VLDB 2001 International Workshop on Databases in Telecommunications II
• Year: 2001

The DMTF's recent work on management information modeling in the IP world has highlighted that a number of problems are still unsolved in this important area of enterprise management In this paper, we identify five: finding the right level of abstraction, building on past experience, devising good models, finding a good trade-off between quality and timeliness of new models, and attracting the best experts in the field in standardization efforts. We propose to alleviate them by splitting information modeling into two phases that involve different people with different skills. In the first phase, designers and experts in a given technology (be it a router, a service, a policy, etc.) capture the core issues for managing it in a Universal Information Model (UIM) tbut is independent of any management architecture. At tiffs stage, low-level engineering details are ignored. In the second phase, code-oriented engineers instantiate the UIM into a data model specific to a management architecture (e.g., an SNMP MIB or a CIM schema). These people are specialists of SNMP or WBEM, but are not necessarily experts in the technology being managed.