# US 12,216,552 B2 · Multi-phase cloud service node error prediction based on minimization function with cost ratio and false positive detection > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,216,552 B2 - Document ID: US12216552B2 - Status: Granted - Kind: B2 - Application number: 17/056,744 - Filed: 2018-06-29 - Published: 2021-07-08 - Granted: 2025-02-04 - Earliest claimed: 2018-06-29 - Assignees: Microsoft Technology Licensing, LLC - Applicants: Microsoft Technology Licensing, LLC - Inventors: Qingwei Lin; Kaixin Sui; Yong Xu - Web page: https://patentgrep.com/patent/US12216552B2 ## CPC - G06F 11/1484: {involving virtual machines} - G06N 20/20: Ensemble learning - G06F 9/45558: {Hypervisor-specific management and integration aspects} - G06F 9/4856: {resumption being on a different machine, e.g. task migration, virtual machine migration (G06F9/5088 takes precedence)} - G06F 9/5072: {Grid computing} - G06F 9/5077: {Logical partitioning of resources; Management or configuration of virtualized resources (specific details on emulation or internal functioning of virtual machines G06F9/455)} - G06F 11/004: {Error avoidance (G06F11/07 and subgroups take precedence)} - G06F 11/008: {Reliability or availability analysis} - G06F 11/142: {Reconfiguring to eliminate the error (group management mechanisms in a peer-to-peer network H04L67/1044)} - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06F 2009/45562: {Creating, deleting, cloning virtual machine instances} - G06F 2009/4557: {Distribution of virtual machine instances; Migration and load balancing} - G06F 2009/45591: {Monitoring or debugging support} - G06F 2201/81: Threshold - G06F 2201/815: Virtual - G06N 3/042: Knowledge-based neural networks; Logical representations of neural networks - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Systems and techniques for multi-phase cloud service node error prediction are described herein. A set of spatial metrics and a set of temporal metrics may be obtained for node devices in a cloud computing platform. The node devices may be evaluated using a spatial machine learning model and a temporal machine learning model to create a spatial output and a temporal output. One or more potentially faulty nodes may be determined based on an evaluation of the spatial output and the temporal output using a ranking model. The one or more potentially faulty nodes may be a subset of the node devices. One or more migration source nodes may be identified from one or more potentially faulty nodes. The one or more migration source nodes may be identified by minimization of a cost of false positive and false negative node detection. ## Claims ### Claim 1 A system for predicting computing node failure in a cloud computing platform, the system comprising: at least one processor; and memory including instructions that, when executed by the at least one processor, cause the at least one processor to perform operations to: obtain a set of spatial metrics and a set of temporal metrics for computing node devices in the cloud computing platform, the set of spatial metrics comprising spatial signals from hardware and software components shared by the computing node devices, and the set of temporal metrics comprising temporal signals from hardware and software components for each computing node device of the computing node devices; evaluate the computing node devices using a spatial machine learning model and the set of spatial metrics and using a temporal machine learning model and the set of temporal metrics to create a spatial output and a temporal output for each computing node device of the computing node devices; determine one or more potentially faulty computing node devices based on an evaluation of the spatial output and the temporal output using a ranking model, wherein the one or more potentially faulty computing node devices is a subset of the computing node devices; identify one or more migration source computing no de devices from the one or more potentially faulty computing node devices, wherein a number of computing node devices included in the one or more migration source computing node devices are determined using a threshold calculated by applying a minimization function to a cost ratio and a predicted number of false positive detections included in the one or more potentially faulty computing node devices and a predicted number of false negative detections excluded from the one or more potentially faulty computing node devices, the cost ratio representing a ratio between a historical cost of false positive detection and a historical cost of false negative detection; identify one or more migration target computing node devices from one or more potentially healthy computing node devices; and migrate a virtual machine (VM) from a faulty computing node device of the one or more migration source computing node devices to a healthy computing node device of the one or more migration target computing node devices. ### Claim 2 (depends on 1) The system of claim 1, wherein the memory further includes instructions to generate the spatial machine learning model using random forest training. ### Claim 3 (depends on 1) The system of claim 1, wherein memory further includes instructions to generate the temporal machine learning model using long short-term memory training. ### Claim 4 (depends on 1) The system of claim 1, wherein the instructions to determine the one or more potentially faulty computing node devices further includes instructions to: obtain a spatial output vector of trees of the spatial machine learning model; obtain a temporal output vector of a dense layer of the temporal machine learning model; concatenate the spatial output vector and the temporal output vector to form an input vector for the ranking model; and generate a ranking of the computing node devices using the ranking model, wherein the one or more potentially faulty computing node devices is a subset of the ranked computing node devices. ### Claim 5 (depends on 1) The system of claim 1, wherein the set of temporal metrics are obtained from respective computing node devices of the cloud computing platform, wherein a computing node device includes a physical computing device that hosts one or more virtual machines (VMs). ### Claim 6 (depends on 1) The system of claim 1, wherein the set of spatial metrics are obtained from a node controller for respective computing node devices of the cloud computing platform. ### Claim 7 (depends on 1) The system of claim 1, wherein the spatial machine learning model is generated using a training set of spatial metrics, and wherein the training set of spatial metrics include metrics shared by two or more respective computing node devices. ### Claim 8 (depends on 1) The system of claim 1, wherein the temporal machine learning model is generated using a training set of temporal metrics, and wherein the training set of temporal metrics include metrics individual to respective computing node devices. ### Claim 9 (depends on 1) The system of claim 1, the memory further including instructions to: identify the one or more potentially healthy computing node devices based on the evaluation of the spatial output and the temporal output using the ranking model, wherein the one or more potentially healthy computing node devices is a subset of the computing node devices. ### Claim 10 (depends on 1) The system of claim 1, the memory further including instructions to: identify the one or more potentially healthy computing node devices based on the evaluation of the spatial output and the temporal output using the ranking model, wherein the one or more potentially healthy computing node devices is a subset of the computing node devices; identify the one or more migration target computing node devices from the one or more potentially healthy computing node devices; and create a new virtual machine (VM) on a healthy computing node device of the one or more migration target computing node devices in lieu of a faulty node of the one or more migration source computing node devices. ### Claim 11 A method for predicting computing node failure in a cloud computing platform, the method comprising: obtaining a set of spatial metrics and a set of temporal metrics for computing node devices in the cloud computing platform, the set of spatial metrics comprising spatial signals from hardware and software components shared by the computing node devices, and the set of temporal metrics comprising temporal signals from hardware and software components for each computing node device of the computing node devices; evaluating the computing node devices using a spatial machine learning model and the set of spatial metrics and using a temporal machine learning model and the set of temporal metrics to create a spatial output and a temporal output for each computing node device of the computing node devices; determining one or more potentially faulty computing node devices based on an evaluation of the spatial output and the temporal output using a ranking model, wherein the one or more potentially faulty computing node devices is a subset of the computing node devices; identifying one or more migration source computing node devices from the one or more potentially faulty computing node devices, wherein a number of computing node devices included in the one or more migration source computing node devices are determined using a threshold calculated by applying a minimization function to a cost ratio and a predicted number of false positive detections included in the one or more potentially faulty computing node devices and a predicted number of false negative detections excluded from the one or more potentially faulty computing node devices, the cost ratio representing a ratio between a historical cost of false positive detection and a historical cost of false negative detection; identifying one or more migration target computing node devices from one or more potentially healthy computing node devices; and migrating a virtual machine (VM) from a faulty computing node device of the one or more migration source computing node devices to a healthy computing node device of the one or more migration target computing node devices. ### Claim 12 (depends on 11) The method of claim 11, wherein determining the one or more potentially faulty computing node devices further comprises: obtaining a spatial output vector of trees of the spatial machine learning model; obtaining a temporal output vector of a dense layer of the temporal machine learning model; concatenating the spatial output vector and the temporal output vector to form an input vector for the ranking model; and generating a ranking of the computing node devices using the ranking model, wherein the one or more potentially faulty computing node devices is a subset of the ranked computing node devices. ### Claim 13 (depends on 11) The method of claim 11, wherein the spatial machine learning model is generated using a training set of spatial metrics, and wherein the training set of spatial metrics include metrics shared by two or more respective computing node devices. ### Claim 14 (depends on 11) The method of claim 11, further comprising: identifying the one or more potentially healthy computing node devices based on the evaluation of the spatial output and the temporal output using the ranking model, wherein the one or more potentially healthy computing node devices is a subset of the computing node devices. ### Claim 15 (depends on 11) The method of claim 11, further comprising: identifying the one or more potentially healthy computing node devices based on the evaluation of the spatial output and the temporal output using the ranking model, wherein the one or more potentially healthy computing node devices is a subset of the computing node devices; identifying the one or more migration target computing node devices from the one or more potentially healthy computing node devices; and creating a new virtual machine (VM) on a healthy computing node device of the one or more migration target computing node devices in lieu of a faulty computing node device of the one or more migration source computing node devices. ### Claim 16 At least one non-transitory machine-readable medium comprising instructions for predicting computing node failure in a cloud computing platform that, when executed by at least one processor, cause the at least one processor to perform operations to: obtain a set of spatial metrics and a set of temporal metrics for computing node devices in the cloud computing platform, the set of spatial metrics comprising spatial signals from hardware and software components shared by the computing node devices, and the set of temporal metrics comprising temporal signals from hardware and software components for each computing node device of the computing node devices; evaluate the computing node devices using a spatial machine learning model and the set of spatial metrics and using a temporal machine learning model and the set of temporal metrics to create a spatial output and a temporal output for each computing node device of the computing node devices; determine one or more potentially faulty computing node devices based on an evaluation of the spatial output and the temporal output using a ranking model, wherein the one or more potentially faulty computing node devices is a subset of the computing node devices; identify one or more migration source computing node devices from the one or more potentially faulty computing node devices, wherein a number of computing node devices included in the one or more migration source computing node devices are determined using a threshold calculated by applying a minimization function to a cost ratio and a predicted number of false positive detections included in the one or more potentially faulty computing node devices and a predicted number of false negative detections excluded from the one or more potentially faulty computing node devices, the cost ratio representing a ratio between a historical cost of false positive detection and a historical cost of false negative detection; identify one or more migration target computing node devices from one or more potentially healthy computing node devices; and migrate a virtual machine (VM) from a faulty computing node device of the one or more migration source computing node devices to a healthy computing node device of the one or more migration target computing node devices. ### Claim 17 (depends on 16) The at least one non-transitory machine-readable medium of claim 16, further comprising instructions to generate the spatial machine learning model using random forest training. ### Claim 18 (depends on 16) The at least one non-transitory machine-readable medium of claim 16, further comprising instructions to generate the temporal machine learning model using long short-term memory training. ### Claim 19 (depends on 16) The at least one non-transitory machine-readable medium of claim 16, wherein the instructions to determine the one or more potentially faulty computing node devices further includes instructions to: obtain a spatial output vector of trees of the spatial machine learning model; obtain a temporal output vector of a dense layer of the temporal machine learning model; concatenate the spatial output vector and the temporal output vector to form an input vector for the ranking model; and generate a ranking of the computing node devices using the ranking model, wherein the one or more potentially faulty computing node devices is a subset of the ranked computing node devices. ### Claim 20 (depends on 16) The at least one non-transitory machine-readable medium of claim 16, wherein the set of temporal metrics are obtained from respective computing nodes of the cloud computing platform, wherein a node includes a physical computing device that hosts one or more virtual machines (VMs).