Virtual machine underutilization detector
AssigneeRoyal Bank of Canada
InventorsMarat KHANDROS, Amirreza OGHBAEE, Ali SYED, Kamran PIRASTEH, Weitao LIN, Alexander Michael KRUSH, Jacob Alexander ABBOTT
Life of this patent
Estimating the term…
Abstract
Systems and methods are provided for detecting sub-optimal performance of one or more virtual computing platforms. Usage data representing user activity, and performance data representing computing hardware resource utilization, is collected from a plurality of virtual machines hosted on one or more virtual computing platforms. The usage data and performance data is then analyzed along with configuration data representing the hardware components of the computing devices operating the virtual computing platform.
Claims20
A computer implemented system for detecting sub-optimal performance of one or more virtual computing platforms, the system comprising:
- a middleware computing node configured to receive usage data and the performance data from one or more data sources;
- translate the usage and the performance data into a unified format;
- store the translated usage and performance data in a memory;
- and transmit the usage and performance data;
- a database computing node configured to receive the usage data and the performance data in the unified format from the middleware node and to store the usage data and the performance data in a data storage;
- and
- a classification computing node configured to process the usage data and the performance data in the unified format in order to produce a utilization score and a classification for the one or more data sources and store the utilization score and the classification in a memory, the classification node comprising:
- a feature detection engine for detecting, collecting, interpreting, and quantifying features of usage data and performance data in the unified format and storing the features in a memory;
- a scoring engine for producing the utilization score for the one or more data sources based on at least one of the features stored in the memory by the feature detection engine, and storing the utilization score in the memory;
- and
- a classification engine assigning the one or more data sources into one or more classes based on the utilization score and at least one of the features of the one or more data sources stored in the memory.
A computer implemented method for detecting sub-optimal performance of one or more virtual computing platforms, the method comprising:
- receiving usage data and performance data associated with the one or more virtual computing platforms;
- detecting, collecting, interpreting, and quantifying features of usage data and performance data and storing the features in a memory;
- producing one or more utilization scores corresponding to each virtual computing platform of the one or more virtual computing platforms based on at least one of the features stored in the memory by the feature detection engine, and storing the utilization score in the memory;
- assigning the one or more virtual computing platforms into one or more classes based on the utilization score;
- wherein the one or more utilization scores are generated using a machine learning computing model, the machine learning computing model including a first expectation maximization step using a Kalman filter initialized with evenly weighted prior Gaussian distributions.
A computer readable medium storing a set of machine-interpretable instructions, which when executed, cause a processor to perform steps of a method for detecting sub-optimal performance of one or more virtual computing platforms, the method comprising:
- receiving usage data and performance data associated with the one or more virtual computing platforms;
- detecting, collecting, interpreting, and quantifying features of usage data and performance data and storing the features in a memory;
- producing one or more utilization scores corresponding to each virtual computing platform of the one or more virtual computing platforms based on at least one of the features stored in the memory by the feature detection engine, and storing the utilization score in the memory;
- assigning the one or more virtual computing platforms into one or more classes based on the utilization score;
- wherein the one or more utilization scores are generated using a machine learning computing model, the machine learning computing model including a first expectation maximization step using a Kalman filter initialized with evenly weighted prior Gaussian distributions.
Dependencies are inferred from the claim text.