Systems and methods for automatically generating a data center network mapping for automated alarm consolidation
Inventors
Healey, Christopher M. • Lake, Anna
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11743104-B2
Publication Date
2023-08-29
Expiration Date
Abstract
According to at least one aspect of the present invention, a system for automatically generating a data-center network mapping for automated alarm consolidation is provided comprising a plurality of devices, and at least one computing device communicatively coupled to each of the devices, the at least one computing device being configured to receive operational data from each of the devices, the operational data being indicative of at least one of a power path, cooling or temperature zones, or communications paths, determine, based on the operational data, device relationships between each of the devices, receive, from each of the devices, a respective alarm of a plurality of alarms, determine alarm relationships between at least two of the plurality of alarms, consolidate, based on the determined device relationships and based on the determined alarm relationships, the plurality of alarms into a consolidated alert, and provide the consolidated alert to a user.
Core Innovation
The invention describes an automated system for automatically generating a data-center network mapping for a plurality of interrelated devices. The system receives operational data from each device, where the operational data is indicative of at least one of a power path, cooling or temperature zones, or communications paths. Based on the operational data, the system determines device relationships between devices in the plurality and determines device types for a first device and a second device.
Using the determined first device type and second device type, the system determines that the first device is related or unrelated to the first device type. The device relationships are inferred from operational data and the mapping is generated automatically for interrelated devices across device relationships and device types. The approach supports identifying relationships among device types rather than only direct device-to-device linkage.
The operational data used for relationship inference includes preprocessing that resamples or normalizes to a single polling frequency, and rounds and interpolates missing measurements. Determining device relationships may use Bayesian network estimation, correlation coefficient analysis with p-values and false discovery rate, restricted regression optimization, and/or ordinary least squares with optional time factors. The inferred relationships are then used to support downstream processing that consolidates multiple device alarms into a consolidated alert.
Claims Coverage
The document includes three independent claims: one system claim, one non-transitory computer-readable medium claim, and one method claim. Each independent claim covers the same core sequence of features: receiving operational data indicative of data-center aspects, determining device relationships, determining device types, and determining whether a device is related or unrelated to a device type, with dependent claims further refining preprocessing and modeling techniques.
Automatically generating a data-center network mapping from operational data
Receive operational data from each device of the plurality of interrelated devices, the operational data being indicative of at least one of a power path, cooling or temperature zones, or communications paths.
Inferring device relationships between interrelated devices
Determine, based on the operational data, device relationships between each device of the plurality of interrelated devices.
Determining device types for first and second devices
Determine, based on the operational data, that a first device of the plurality of interrelated devices is a first device type and that a second device of the plurality of interrelated devices is a second device type.
Determining related or unrelated based on device types
Determine, based on determining that the first device is a first device type and that the second device is a second device type, that the first device is related or unrelated to the first device type.
Applying device relationship inference with standardized operational data
Standardize operational data using a single polling frequency when determining device relationships.
Using rounding to a polling-period boundary and interpolation
Round data measurements to the nearest period of a single polling frequency and interpolate data measurements for each period of a single polling frequency where no data measurement exists.
Generating device mappings using specified estimation/regression techniques
Use one or more of Bayesian network estimation, correlation coefficient techniques, restricted regression optimization techniques, and ordinary least squares techniques to generate a device mapping indicating relationships among devices in a plurality of interrelated devices.
Including power-device measurement data as operational data
Include operational data that includes current data, power data, voltage data, and/or temperature data for a power device.
Across the independent system, medium, and method claims, the core inventive coverage is automatically generating a data-center network mapping by receiving operational data, determining device relationships, determining device types, and determining whether devices are related or unrelated based on device types. Dependent claims further specify operational-data standardization, named relationship-estimation/regression techniques, and inclusion of power-device current, power, voltage, and temperature data.
Stated Advantages
Consolidates multiple device alarms into a single consolidated alert for a user.
Supports downstream uses of the device mapping such as alarm filtering, reliability/health estimation, and impact/replacement recommendations.
Documented Applications
Alarm consolidation for a user by grouping alarms into subsets and using derived association rules to produce consolidated alerts.
Downstream alarm filtering using the inferred device mapping.
Reliability/health estimation using the inferred device mapping and inferred alarm relationships.
Impact/replacement recommendations using the inferred device mapping and inferred device relationships.
Interested in licensing this patent?