Detection of underutilized data center resources

A method for identification of an underutilized resource in a data center, the method comprising:

• obtaining, using a computer processor, historical telemetry data and historical trace log data related to three or more performance metrics from each resource of a resource pair from two or more resource pairs spanning three or more months of usage from more than six months before a current date;
• wherein:
• a resource comprises a server, a hypervisor, or a virtual machine;
• the resource pair comprises an active server and an idle server, an active hypervisor and an idle hypervisor, or an active virtual machine and an idle virtual machine;
• the historical telemetry data come from measurements made by one or more telemetry agents located in proximity to the resource pair which they measure;
• the historical trace log data come from measurements made by one or more trace log agents located in proximity to the resource pairs which they measure;
• a performance metric comprises electricity usage, temperature data, input/output (I/O) utilization, inbound network activity related to secondary and tertiary applications, central processing unit (CPU) utilization, memory usage, connections, disk usage, and fan speed;
• calculating, using the computer processor, activity levels of the resources located in one or more data centers using the historical telemetry data and historical trace log data;
• training, using the computer processor, an artificial intelligence/machine learning (“AI/ML”) model, using the historical telemetry data, the historical trace log data, and calculated activity levels, to predict an activity level of a resource based on recent telemetry data and recent trace log data;
• wherein:
• the activity level is a ratio of an average usage of a resource over a period to a maximum usage potential of the resource;
• obtaining measurements, using the computer processor, of performance metrics of a resource in a data center by obtaining recent telemetry data and recent trace log data spanning three or more months within six months before the current date;
• wherein:
• the recent telemetry data is measured with a telemetry agent in proximity to the resource;
• the recent trace log data is measured with a trace log agent in proximity to the resource;
• determining, by running the AI/ML model on the computer processor, the activity level of the resource over a time spanning three or more months within the most recent six months;
• wherein:
• to be designated as an active server, an active hypervisor, and an active virtual machine, the activity level is at least 10%;
• to be designated as an idle server, an idle hypervisor, and an idle virtual machine, the activity level is less than 10%;
• when the activity level of the resource is less than 10% spanning three or more months within six months before the current date, running, using the computer processor:
• a status check engine to assess applications currently running on the resource are part of routine maintenance;
• and
• a discovery engine to assess applications running on the resource spanning months, when a status of idle was designated by the AI/ML model, were part of routine maintenance;
• when the status of idle was designated by the AI/ML model, using the computer processor to run an analytics engine to verify that applications assessed to be part of routine maintenance by the AI/ML model are indeed part of routine maintenance based on assessing how the applications have been used on the resource over a lifetime of use of the resource in the data center;
• and
• when verifying that the applications are part of routine maintenance, notifying an administrator of the data center that the resource is idle.

An apparatus for automating identification of an underutilized resource in a data center, the apparatus comprising:

• a computer processor running in a data center;
• an artificial intelligence/machine learning (“AI/ML”) model running on the computer processor;
• one or more resources located in the data center, wherein the resource comprises a server, a hypervisor, and a virtual machine;
• two or more resource pairs, wherein the resource pair comprises an active server and an idle server, an active hypervisor and an idle hypervisor, or an active virtual machine and an idle virtual machine;
• an application in communication with the resource;
• a status check engine operating in the data center;
• a discovery engine operating in the data center;
• an analytics engine operating in the data center;
• a trace log agent located in proximity to the resource and in communication with the resource and the application;
• a telemetry agent located in proximity to the resource and in communication with the resource;
• wherein the computer processor is configured to:
• use the trace log agent to obtain historical trace log data from the one or more resources of the two or more resource pairs spanning three or more months of usage from more than six months before a current date;
• use the telemetry agent to obtain historical telemetry data related to three or more performance metrics from the one or more resources of the two or more resource pairs spanning three or more months of usage from more than six months before a current date;
• wherein the performance metric comprises electricity usage, temperature data, input/output (I/O) utilization, inbound network activity related to secondary and tertiary applications, central processing unit (CPU) utilization, memory usage, connections, disk usage, and fan speed;
• calculate activity levels of the one or more resources located in the data center using the historical telemetry data and historical trace log data;
• wherein the performance metric comprises electricity usage, temperature data, input/output (I/O) utilization, inbound network activity related to secondary and tertiary applications, central processing unit (CPU) utilization, memory usage, connections, disk usage, and fan speed;
• train the AI/ML model, using the historical telemetry data, the historical trace log data, and calculated activity levels, to predict a recent activity level of the one or more resources based on recent telemetry data and recent trace log data;
• wherein the activity level is a ratio of an average usage of a resource over a period to a maximum usage potential of the resource;
• obtain measurements of performance metrics of a resource in a data center by obtaining recent telemetry data and recent trace log data spanning three or more months within six months before the current date;
• determine, by running the AI/ML model, the activity level of the resource over a time spanning three or more months within the most recent six months;
• wherein:
• to be designated as an active server, an active hypervisor, and an active virtual machine, the activity level is at least 10%;
• and
• to be designated as an idle server, an idle hypervisor, and an idle virtual machine, the activity level is less than 10%;
• run, when the activity level of the resource is less than 10% spanning three or more months within six months before the current date:
• a status check engine to assess applications currently running on the resource are part of routine maintenance;
• and
• a discovery engine to assess applications running on the resource spanning months, when a status of idle was designated by the AI/ML model, were part of routine maintenance;
• run, when a status of idle was designated by the AI/ML model, the analytics engine to verify that applications assessed to be part of routine maintenance by the AI/ML model are indeed part of routine maintenance based on assessing how the applications have been used on the resource over a lifetime of use of the resource in the data center;
• and
• notify, when verifying that the applications are part of routine maintenance, an administrator of the data center that the resource is idle.