摘要 |
The current document discloses an automated method and system for inferring the logical rules underlying the configuration and versioning state of the components and subcomponents of a complex system, including data centers and other complex computational environments. The methods and systems employ a database of configuration information and construct an initial set of logical rules, or hypotheses, regarding system configuration. Then, using simulated annealing and a variant of genetic programming, the methods and systems disclosed in the current document carry out a search through the hypothesis state space for the system under several constrains in order to find one or more hypotheses that best explain the configuration and, when available, configuration history. The constraints include minimization of the complexity of the hypotheses and maximizing the accuracy by which the hypotheses predict observed configuration and configuration history. |
主权项 |
1. A system that generates a configuration-hypothesis for a subject complex system that includes a number of components, the system comprising:
one or more processors, one or more data-storage devices, including an electronic memory; observed configuration data for the subject complex system stored in one or more of the one or more data-storage devices; and computer instructions stored in one or more of the one or more data-storage devices that control the system to
generate a set of initial candidate hypotheses and store the initial candidate hypotheses in a data-storage device as a current set of candidate hypotheses, each hypothesis representing a logic-based explanation of the observed configuration data for a number of target components less than or equal to the number of components,iteratively
generate a new set of candidate hypotheses from the current set of hypotheses,select, from the new set of candidate hypotheses, a set of selected hypotheses based on a score assigned by a scoring function to each hypotheses in the new set of candidate hypotheses, andstore the selected hypotheses in the data-storage device as the current set of candidate hypotheses,select, based on a score assigned by a scoring function to each hypotheses in the current set of hypotheses, a final hypothesis that describes the observed configuration data, andstore the selected final hypothesis in one of the one or more data-storage devices. |