Monitoring and Management Services for Distributed Object Systems

As computer architectures evolve away from monolithic applications towards distributed, cooperating sets of objects of higher and higher granularity it becomes increasingly vital to have sophisticated distributed monitoring and management services. Systems containing tens of thousands of objects deployed over thousands of nodes will be deployed. The monitoring and management requirements for different groups of objects will vary widely. Traditional approaches to monitoring and management are inadequate to address these widely varying needs as they typically operate at the level of a platform node or a system rather than the object process. Ongoing work in the CORBA community and OMG CORBA Facilites Task Force will provide a basis for system-level management service for this proposed work. The innovation of the Phase I approach is that individual objects are responsible for certain aspects of their own monitoring and management. These objects possess behaviors which can be invoked from well-defined interfaces. The objects will be able to delegate certain functions to system-level management services, which would possess the breadth of information necessary to make appropriate decisions and take action. This is a multi-level management model where the granularity of the architecture matches exactly the requisite granularity for large-scale object systems.