Management system for provisioning server resources of a data center
AssigneeSuper Micro Computer, Inc.
InventorsCheng-Yi Fang, Chih Chia Huang
Life of this patent
Estimating the term…
Abstract
A data center has a management system for selecting a server computer to start or shut down. The management system has a machine learning model that is trained to predict power consumption of the data center using a large training dataset of many, different data centers. The machine learning model is fine-tuned using data of server computers of the data center. Input data that include temperature information of a server computer and position of the server computer are input to the machine learning model to obtain a predicted difference in power consumption of the data center. Predicted differences in power consumption of the data center are compared to select a server computer to start or shut down.
Claims10
A method of providing a server resource in a data center, the method comprising:
- training a machine learning model using an initial training dataset comprising temperature information, server positions, and power consumption information of server computers of different data centers;
- fine-tuning the machine learning model using fine-tuning data comprising temperature information, server positions, and power consumption information of a plurality of server computers of the data center;
- after the machine-learning model has been fine-tuned, sending prediction requests to the machine learning model for server computers of a subset of the plurality of server computers that are powered OFF, each of the prediction requests including at least a position in the data center of a respective server computer that is powered OFF;
- for each of the prediction requests, using the machine learning model to generate a predicted difference in power consumption of the data center;
- comparing predicted differences in power consumption of the data center to identify a selected server computer among the subset of the plurality of server computers that are powered OFF but when powered ON will result in a lowest power consumption of the data center relative to powering ON other server computers of the subset;
- and
- starting the selected server computer by powering ON the selected server computer responsive to receiving a request for a server resource.
A computer system comprising at least one processor and a memory, the memory storing instructions that when executed by the at least one processor cause the computer system to:
- train a machine learning model to predict power consumption of a data center using an initial training dataset comprising temperature information, server positions, and power consumption information of server computers of a plurality of different data centers;
- fine-tune the machine learning model using fine-tuning data comprising temperature information, server positions, and power consumption information of a plurality of server computers of the data center;
- after the machine learning model is fine-tuned, send prediction requests to the machine learning model for server computers of a subset of the plurality of server computers that are powered OFF, each of the prediction requests including at least a position in the data center of a respective server computer that is powered OFF;
- for each of the prediction requests, use the machine learning model to generate a predicted difference in power consumption of the data center;
- compare predicted differences in power consumption of the data center to identify a selected server computer among the subset of the plurality of server computers that s are powered OFF but when powered ON will result in a lowest power consumption of the data center relative to powering ON other server computers of the subset;
- and
- start the selected server computer by powering ON the selected server computer responsive to receiving a request for a server resource.
Dependencies are inferred from the claim text.