# US 12,182,616 B2 · Platform health engine in infrastructure processing unit > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,182,616 B2 - Document ID: US12182616B2 - Status: Granted - Kind: B2 - Application number: 17/484,099 - Filed: 2021-09-24 - Published: 2022-01-20 - Granted: 2024-12-31 - Earliest claimed: 2021-09-24 - Assignees: Intel Corporation - Applicants: Intel Corporation - Inventors: Susanne M. Balle; Yamini Nimmagadda; Olugbemisola Oniyinde - Web page: https://patentgrep.com/patent/US12182616B2 ## CPC - G06F 9/4856: {resumption being on a different machine, e.g. task migration, virtual machine migration (G06F9/5088 takes precedence)} - G06F 11/3006: {where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems (multiprogramming arrangements G06F9/46; allocation of resources G06F9/50)} - G06F 9/4881: {Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues} - G06F 9/5088: {involving task migration} - G06F 11/004: {Error avoidance (G06F11/07 and subgroups take precedence)} - G06F 11/2023: {Failover techniques} - G06F 11/203: {using migration} - G06F 11/3024: {where the computing system component is a central processing unit [CPU]} - G06F 11/3058: {Monitoring arrangements for monitoring environmental properties or parameters of the computing system or of the computing system component, e.g. monitoring of power, currents, temperature, humidity, position, vibrations (thermal management in cooling arrangements of a computing system G06F1/206)} - G06F 11/3089: {Monitoring arrangements determined by the means or processing involved in sensing the monitored data, e.g. interfaces, connectors, sensors, probes, agents (software debugging using additional hardware using a specific debug interface G06F11/3656; performance evaluation by tracing or monitoring G06F11/3466)} - G06F 11/3409: {for performance assessment} - G06F 11/3447: {Performance evaluation by modeling} - G06F 11/3476: {Data logging (G06F11/14, G06F11/2205 take precedence)} - G06F 11/3495: {for systems} - G06N 20/00: Machine learning - G06F 11/008: {Reliability or availability analysis} - G06F 2209/5019: Workload prediction - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G06N 3/045: Combinations of networks - G06N 3/084: Backpropagation, e.g. using gradient descent - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A platform health engine for autonomous self-healing in platforms served by an Infrastructure Processing Unit (IPU), including: an analysis processor configured to apply analytics to telemetry data received from a telemetry agent of a monitored platform managed by the IPU, and to generate relevant platform health data; a prediction processor configured to predict, based on the relevant platform health data, a future health status of the monitored platform; and a dispatch processor configured to dispatch a workload of the monitored platform to another platform managed if the predicted future health status of the monitored platform is failure. ## Claims ### Claim 1 A platform health engine for autonomous self-healing in platforms served by an Infrastructure Processing Unit (IPU), comprising: an analysis processor configured to apply analytics to telemetry data received from a telemetry agent of a monitored platform managed by the IPU, and to generate relevant platform health data; a prediction processor configured to predict, based on the relevant platform health data, a future health status of the monitored platform; a dispatch processor configured to dispatch a workload of the monitored platform to another platform if the predicted future health status of the monitored platform is failure; and a collection processor configured to collect, and provide to the analysis processor, the telemetry data from the telemetry agent of the monitored platform, wherein if the predicted future health status of the monitored platform is unclear or failure, the prediction processor is configured to request the collection processor to provide the analysis processor with additional telemetry data of the monitored platform. ### Claim 2 (depends on 1) The platform health engine of claim 1, wherein the prediction processor is configured to predict, more accurately based on the additional telemetry data, the future health status of the monitored platform. ### Claim 3 (depends on 1) The platform health engine of claim 1, wherein the collection processor is configured to collect the telemetry data from the telemetry agent of the monitored platform periodically. ### Claim 4 (depends on 1) The platform health engine of claim 1, wherein the collection processor is configured to collect the telemetry data from the telemetry agent of the monitored platform upon a triggering event. ### Claim 5 (depends on 4) The platform health engine of claim 4, wherein the triggering event is a platform failure or a request from the prediction processor. ### Claim 6 (depends on 1) The platform health engine of claim 1, wherein the dispatch processor is configured to dispatch the workload of the monitored platform to the other platform by replicating the workload of the monitored platform on the other platform while the monitored platform continues to run the workload. ### Claim 7 (depends on 1) The platform health engine of claim 1, wherein if the predicted future health status of the monitored platform is non-failure, the dispatch processor is configured to continue to dispatch the workload of the monitored platform to the monitored platform. ### Claim 8 (depends on 1) The platform health engine of claim 1, wherein the prediction processor is configured to provide the IPU with a recommended self-healing action based on the relevant platform health data. ### Claim 9 (depends on 1) The platform health engine of claim 1, wherein the prediction processor is configured to predict the future health status of the monitored platform using a machine learning algorithm. ### Claim 10 (depends on 9) The platform health engine of claim 9, wherein the prediction processor is configured to host a local machine learning model of the machine learning algorithm. ### Claim 11 (depends on 10) The platform health engine of claim 10, wherein a central machine learning model is hosted in a data center server that serves the IPU, and used to update weights of the local machine learning model. ### Claim 12 (depends on 11) The platform health engine of claim 11, wherein the prediction processor is configured to receive from the data center server data from another IPU served by the data center server. ### Claim 13 (depends on 1) The platform health engine of claim 1, wherein the analysis processor comprises one or more filters configured to filter out non-relevant data from the telemetry data, and convert the relevant platform health data to a format readable by the prediction processor. ### Claim 14 (depends on 1) The platform health engine of claim 1, wherein the monitored platform is an accelerator. ### Claim 15 (depends on 1) The platform health engine of claim 1, wherein the platform health data is selected from the group consisting of thermal data, vibration data, utilization data, and memory errors. ### Claim 16 (depends on 1) A data center system comprising: an Infrastructure Processing Unit (IPU); and the platform health engine according to claim 1. ### Claim 17 A platform health engine for autonomous self-healing in platforms served by an Infrastructure Processing Unit (IPU), comprising: an analyzing means for applying analytics to telemetry data received from a telemetry agent of a monitored platform managed by the IPU, and for generating relevant platform health data; a predicting means for predicting, based on the relevant platform health data, a future health status of the monitored platform; and a dispatching means for dispatching a workload of the monitored platform to another platform if the predicted future health status of the monitored platform is failure by replicating the workload of the monitored platform on the other platform while the monitored platform continues to run the workload. ### Claim 18 (depends on 17) The platform health engine of claim 17, further comprising: a collecting means for collecting, and for providing to the analyzing means, the telemetry data from the telemetry agent of the monitored platform. ### Claim 19 (depends on 17) The platform health engine of claim 17, wherein the monitored platform is an accelerator. ### Claim 20 A method for autonomous self-healing in platforms served by an Infrastructure Processing Unit (IPU), comprising: applying, by an analysis processor, analytics to telemetry data received from a telemetry agent of a monitored platform managed by the IPU; generating, by the analysis processor, relevant platform health data; predicting, by a prediction processor based on the relevant platform health data, a future health status of the monitored platform; and dispatching, by a dispatch processor, a workload of the monitored platform to another platform if the predicted future health status of the monitored platform is failure by replicating the workload of the monitored platform on the other platform while the monitored platform continues to run the workload. ### Claim 21 (depends on 20) A non-transient computer-readable storage medium storing instruction which, when executed by a processor, implement the method for autonomous self-healing in platforms served by the IPU according to claim 20.