|
Abstract : |
To a growing degree, applications are expected to be self-configuring and self-managing, and as the range of permissible configurations grows, this is becoming an enormously complex undertaking. Indeed, the management subsystem for a distributed system is often more complex than the application itself. Yet the technology options for building management mechanisms have lagged. Current solutions, such as cluster management systems, directory services, and event notification services, either do not scale adequately or are designed for relatively static settings. In this paper, we describe a new information management service called Astrolabe. Astrolabe monitors the dynamically changing state of a collection of distributed resources, reporting summaries of this information to its users. Like DNS, Astrolabe organizes the resources into a hierarchy of domains, which we call zones to avoid confusion, and associates attributes with each zone. Unlike DNS, the attributes may be highly dynamic, and updates propagate quickly; typically, in tens of seconds. Astrolabe continuously computes summaries of the data in the system using on-the-fly aggregation. The aggregation mechanism is controlled by SQL queries, and can be understood as a type of data mining capability. For example, Astrolabe aggregation can be used to monitor the status of a set of servers scattered within the network, to locate a desired resource on the basis of its attribute values, or to com-, |
