DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is responsive to the Applicant’s amendments filed on 12/10/2025. Claims 1-20 remain pending in the application. Claims 1-2, 14-15, and 18-19 have been amended. Any examiner’s note, objection, and rejection not repeated is withdrawn due to Applicant’s amendment.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 03/29/2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Examiner’s Note
The Examiner cites particular columns, paragraphs, figures, and line numbers in the references as applied to the claims below for the convenience of the applicant. Although the specified citations are representative of the teachings in the art and are applied to the specific limitations within the individual claim, other passages and figures may also apply. It is respectfully requested that, in preparing responses, the Applicant fully consider the references in its entirety as potentially teaching all or part of the claimed invention, as well as the context of the passage as taught by the prior art or disclosed by the Examiner.
Claim Objections
Claims 3, 16, and 20 are objected to because of the following informalities: the use of “/” in the element “server traffic/demand data” should be replaced with appropriate wording to convey the intended meaning. The examiner suggests “server traffic or demand data”. For the purposes of examination, the examiner interprets the element as “server traffic or demand data”.
Appropriate correction is required.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 5, 7-9, 14-15, and 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Vinson et al. (US 20230305901 A1) hereafter Vinson in view of Bulut et al. (US 20210055933 A1) hereafter Bulut, further in view of Wu et al. (US 20150180719 A1) hereafter Wu, further in view of Conley (US 10852705 B1).
Regarding claim 1, Vinson teaches:
A method for a machine learning based scheduler to recommend resource management of a computing system at run time, comprising:
generating training data from a retrospective analysis of historical resource management data associated with the computing system by: (Paragraph 79; “training data (e.g., data relating to historical compute resource usage and scheduling) can be pre-processed to generate feature vectors reflecting the values of various features in the training data. The feature vectors can be used to train the scheduling ML model”, where the training data used to generate feature vectors which are used as training data for the ML model corresponds to generating training data. The data “relating to historical compute resource usage and scheduling” corresponds to being a retroactive analysis of historical resource management data associated with the computing system);
training a machine learning model to optimize resource management of the computing system at run time using the training data (Paragraph 79; “In an embodiment, the scheduling ML model can be trained using suitable training data” corresponds to training a ML model. Optimizing scheduling of resources corresponds to optimizing resource management of the computing system);
Vinson does not explicitly teach generating training data, or the optimization occurring at run time.
However, Bulut teaches:
generating training data (Paragraph 108; “In some embodiments, policy analyzer component 108 can collect (e.g., from database 304) and/or use such operational data as training data to actively learn (e.g., as described above with reference to FIGS. 1, 2, and 3) to identify one or more dependency relationships and/or corresponding dependency relationship direction(s) between a subsequently received new compliance policy and existing compliance policies” corresponds to an explicit generation of training data from operational data).
Vinson and Bulut are considered to be analogous to the claimed invention because they are in the same field of resource scheduling. Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Vinson and Bulut to generate training data from historical data. A person of ordinary skill in the art would have been motivated by the need to create new feature vectors for a resource scheduling machine learning model to operate properly.
Vinson in view of Bulut does not teach that the optimization occurs at run time.
However, Wu teaches:
collecting a set of historical data features from a plurality of servers associated with the computing system (Paragraphs 38, 42; “For example, the historical resource usage data can be stored in the database table 635 to enable the centralized controller 605 to retrieve historical resource usage data for use in determining the optimal number of active servers.”, the retrieval step corresponding to collecting historical data features relating to a plurality of servers associated with the computing system.);
analyzing the set of historical data features (Paragraph 38; “decision engine 615 utilizes the resource utilization information from the active servers in the cluster and historical information to predict a future state of the system and use that prediction to determine whether to turn up more servers or turn down more servers”, where historical information corresponds to the applicant’s set of historical data features.) to determine sleep state solutions for each server in the plurality of servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where assigning servers to standby or sleep states corresponds to determining the operational state, the sleep state solution, for each server in the server pool.) based on a transition delay (Paragraph 40; “decision engine 615 can also determine how many of the inactive servers should be placed in hot standby. The buffer size for the number of servers in hot standby can be determined by the decision engine 615 based on the current trend of workload changes. In some embodiments, the decision engine 615 can estimate the total load (RPS) based on an estimate of time it takes for a server in cold standby to transition into hot standby”, where the decision engine estimates transition time and utilizes that estimate to determine server sleep state/activation decisions.) and power characteristics for each server in the plurality of servers (Paragraph 40; historical resource usage data can be stored in the database table 635 to enable the centralized controller 605 to retrieve historical resource usage data for use in determining the optimal number of active servers... ” which includes “power consumption data”, corresponding to the applicant’s power characteristics. Power characteristics is further disclosed in Paragraph 34; “average power consumption by the servers in the cluster are examples of feedback 520 collected by the centralized controller 505. The power 525 is drawn by the servers 1-N from the power supply 515 and can be used to determine the efficiency of the cluster during a time period”), wherein the sleep state solutions comprise a sleep state recommendation for the plurality of servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where assigning servers to standby or sleep states corresponds to determining the operational state, the sleep state solution, for each server in the server pool.);
optimization occurs at run time (Paragraph 16; “the centralized controller deployed in a cluster continuously maintains just the right amount of server capacity in the cluster to adapt to time-varying workloads (e.g., workload surges or drops) and changing application and system behaviors (e.g., caused by software or hardware changes), and in doing so, optimizes both the latency and efficiency characteristics of the cluster.” corresponds to the goal of optimizing resource management using the training data at run time, as the controller performs optimization based on the behavior during run time of the application.).
Vinson, Bulut, and Wu are considered to be analogous to the claimed invention because they are in the same field of resource scheduling. Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Vinson in view of Bulut to incorporate the teachings of Wu determine server sleep state configurations based on transition delay and power characteristics from analyzing historical data. A person of ordinary skill in the art would have recognized the known method of analyzing historical system information to determine configuration states applied to the system of Vinson in view of Bulut would yield the predictable result of efficiently managing server activation and standby states to reduce power consumption while maintaining system performance.
While Wu teaches the generation of a single sleep state solution for a plurality of servers providing a single recommendation for each server in the plurality of servers, Vinson in view of Bulut, further in view of Wu does not teach wherein the solutions comprise a plurality of recommendations.
However, Conley teaches:
wherein the solutions comprise a plurality of recommendations (Col. 11, line 59 – Col. 12, line 10; “user interface may present one or more notifications or recommendations”, “a recommendation may indicate that a particular electrical device 102 is not normally used between the hours of 12:00 P.M. and 12:00 A.M. and suggest that the electrical device 102 be transitioned to a low power state”, where “[t]he notification may indicate a quantity of electrical power or a financial cost that may be conserved by placing particular electrical devices 102 in a low power state during particular time periods.”, the one or more recommendations thereby contemplating a plurality of recommendations.).
Vinson, Bulut, Wu, and Conley are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut, further in view of Wu to incorporate the teachings of Conley and extend the functionality of the system of Vinson in view of Bulut further in view of Wu to generate a plurality of sleep state solutions instead of a singular one. A person of ordinary skill in the art would have recognized the use of the known method of generating multiple possible solutions would yield the predictable result of providing multiple possible sleep state solutions for the plurality of servers, thereby enabling more flexible management of server power states and system performance.
Claim 14 recites similar limitations as those of claim 1, additionally reciting a processor and a non-transitory computer-readable storage medium. Vinson teaches:
a processor; and a non-transitory computer-readable storage medium (Paragraph 24; “A compute scheduling controller 200 includes a processor 202, a memory 210, and network components 220. The memory 210 may take the form of any non-transitory computer-readable medium”).
Claim 14 is rejected for similar reasons as those of claim 1.
Claim 18 recites similar limitations as those of claim 1, additionally reciting a non-transitory computer-readable storage medium. Vinson teaches:
a non-transitory computer-readable storage medium (Paragraph 24; “A compute scheduling controller 200 includes a processor 202, a memory 210, and network components 220. The memory 210 may take the form of any non-transitory computer-readable medium”).
Claim 18 is rejected for similar reasons as those of claim 1.
Regarding claim 2, Vinson in view of Bulut, further in view of Wu, further in view of Conley teaches the method of claim 1. Wu teaches:
wherein generating the training data from the retrospective analysis of historical resource management data further comprises:
generating the sleep state solution for managing sleep states for the plurality of servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Conley teaches:
the solutions comprise a plurality of recommendations (Col. 11, line 59 – Col. 12, line 10; “user interface may present one or more notifications or recommendations”, “a recommendation may indicate that a particular electrical device 102 is not normally used between the hours of 12:00 P.M. and 12:00 A.M. and suggest that the electrical device 102 be transitioned to a low power state”, where “[t]he notification may indicate a quantity of electrical power or a financial cost that may be conserved by placing particular electrical devices 102 in a low power state during particular time periods.”, the one or more recommendations thereby contemplating a plurality of recommendations.).
Vinson teaches:
used as the training data for the machine learning model (Paragraph 81; “In an embodiment, the scheduling ML model 124 can be initially trained using general training data, and can be updated based on actual usage data for the compute resources” corresponds to utilizing the actual usage data as training data for the machine learning model.).
Claim 15 recites similar limitations as those of claim 2. Claim 15 is rejected for similar reasons as those of claim 2.
Regarding claim 5, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 2. Wu teaches:
the sleep state solution (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Conley teaches:
a server availability schedule (Col. 11, line 59 – Col. 12, line 10; “user interface may present one or more notifications or recommendations”, “a recommendation may indicate that a particular electrical device 102 is not normally used between the hours of 12:00 P.M. and 12:00 A.M. and suggest that the electrical device 102 be transitioned to a low power state”, where “[t]he notification may indicate a quantity of electrical power or a financial cost that may be conserved by placing particular electrical devices 102 in a low power state during particular time periods.”, the particular time periods corresponding to an availability schedule.).
Regarding claim 8, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 1. Vinson teaches:
collecting a second set of training data based, in part, on the implemented optimization recommendations to manage the current state of the computer system (Paragraph 81; “In an embodiment, the scheduling ML model 124 can be initially trained using general training data, and can be updated based on actual usage data for the compute resources (e.g., user overrides and user activity)” corresponds to utilizing the general training data, corresponding to the initial solution which is the initial implemented optimization recommendations to manage the current state of the system. Paragraph 48 further discloses “user override or user activity… occurred within the threshold period” in response to scheduling, and this is utilized as “additional training data”, corresponding to the second set of training data based in part on the implemented optimization recommendations);
and retraining, using the second set of training data, the machine learning model to optimize resource management of the computing system (Paragraphs 47 and 48; “the compute scheduling service re-trains… the ML model” and “use[s] the override or activity as additional training data to update the trained ML Model 124” corresponds to retraining using the second set of training data to optimize resource management of the computing system).
Regarding claim 9, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 1. Vinson teaches:
wherein the optimization recommendations include an optimized solution, wherein the optimized solution comprises a plurality of recommendations for the servers to be applied at run time for a plurality of servers of the computing system (Paragraphs 86 and 89; “schedule recommendation 732” corresponds to a server availability schedule which determines what servers should be available. “schedule recommendation 732 can indicate schedules by region, to allow a compute scheduling service to scale compute resources on a region-by-region basis” and “the ML scheduling service can decline to replace an existing schedule… with an updated schedule if the confidence score is below a threshold” which shows that scheduling recommendations are not static. They must be dynamically applied at runtime because the scheduling service actively evaluates conditions and applies or declines the updated schedule in real time, directly controlling server availability).
Wu teaches:
the sleep state solution for servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Conley teaches:
An availability schedule (Col. 11, line 59 – Col. 12, line 10; “user interface may present one or more notifications or recommendations”, “a recommendation may indicate that a particular electrical device 102 is not normally used between the hours of 12:00 P.M. and 12:00 A.M. and suggest that the electrical device 102 be transitioned to a low power state”, where “[t]he notification may indicate a quantity of electrical power or a financial cost that may be conserved by placing particular electrical devices 102 in a low power state during particular time periods.”, the particular time periods corresponding to an availability schedule.).
Claim 17 recites similar limitations as those of claim 9. Claim 17 is rejected for similar reasons as those of claim 9.
Claims 3, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Bhimani et al. (US 20230196234 A1) hereafter Bhimani, further in view of Sethi et al. (US 20230214309 A1) hereafter Sethi, further in view of Tamir et al. (US 11212195 B1) hereafter Tamir.
Regarding claim 3, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 2. Vinson in view of Bulut further in view of Wu, further in view of Conley does not teach wherein the historical data features are chosen from a group of data features consisting of: system state data, server characteristics, carbon intensity data, and server traffic/demand data.
However, Bhimani teaches:
carbon intensity data (Paragraph 39; “Input features can be extracted from the collected raw data and then used by the prediction models as input to generate individual and/or overall predictions, estimations or forecasts of electricity demands and/or emissions (e.g., carbon intensities, etc.) for some or all of the electricity grids and the electric vehicles for any given time point or interval.”).
Vinson, Bulut, Wu, Conley, and Bhimani are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley to incorporate the teachings of Bhimani and utilize carbon intensity data as a data feature. A person of ordinary skill in the art would recognize that data centers are a significant source of carbon output due to massive energy consumption, and would be motivated by the need to minimize the carbon footprint of data centers.
Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Bhimani does not teach system state data, server characteristics, or server traffic/demand data.
However, Sethi teaches:
system state data and characteristics (Paragraph 37; “In addition to the characteristics of the system state information noted herein above, the system state information comprises data identifying a plurality of factors associated with the collected system state information”).
Vinson, Bulut, Wu, Conley, Bhimani, and Sethi are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley further in view of Bhimani to incorporate the teachings of Sethi and utilize system state data and characteristics as data features. A person of ordinary skill in the art would have been motivated by the need to monitor server metrics for critical servers.
Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Bhimani, further in view of Sethi does not teach server traffic/demand.
However, Tamir teaches:
server traffic/demand (Col. 9, lines 41-44; “the metric thresholds 228 may be configured to increase or decrease based on changes in server processing demand, such as during different times of day (e.g., lower thresholds at night than during the days)”).
Vinson, Bulut, Wu, Conley, Bhimani, Sethi, and Tamir are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley further in view of Bhimani further in view of Sethi to incorporate the teachings of Tamir and have server traffic/demand as a data feature, and have the characteristics taught by Sethi apply to servers. A person of ordinary skill would have been motivated by the need to monitor server metrics for critical servers.
Claim 16 recites similar limitations as those of claim 3. Claim 16 is rejected for similar reasons as those of claim 3.
Claim 20 recites similar limitations as those of claim 3. Claim 20 is rejected for similar reasons as those of claim 3.
Claims 4 and 7 are rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Bhimani.
Regarding claim 4, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 2.
Wu teaches:
a plurality of servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”);
the sleep state solution (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Vinson in view of Bulut further in view of Wu further in view of Conley does not teach based on a carbon intensity forecast.
However, Bhimani teaches:
based on a carbon intensity forecast (Paragraph 39; “Input features can be extracted from the collected raw data and then used by the prediction models as input to generate individual and/or overall predictions, estimations or forecasts of electricity demands and/or emissions (e.g., carbon intensities, etc.) for some or all of the electricity grids and the electric vehicles for any given time point or interval.”).
Vinson, Bulut, Wu, Conley, and Bhimani are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley to incorporate the teachings of Bhimani and utilize carbon intensity data as a data feature. A person of ordinary skill in the art would recognize that data centers are a significant source of carbon output due to massive energy consumption, and would be motivated by the need to minimize the carbon footprint of data centers.
Regarding claim 7, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 1. Vinson teaches:
related to the computing system (Paragraph 32; “FIG. 3 illustrates a scheduler workflow 300 for automatic compute environment scheduling using machine learning, according to an embodiment.”, where the compute environment corresponds to the computing system).
Vinson in view of Bulut further in view of Wu further in view of Conley does not teach wherein the optimization recommendations are configured to reduce carbon intensity values.
However, Bhimani teaches:
wherein the optimization recommendations are configured to reduce carbon intensity values (Paragraph 39; “Input features can be extracted from the collected raw data and then used by the prediction models as input to generate individual and/or overall predictions, estimations or forecasts of electricity demands and/or emissions (e.g., carbon intensities, etc.) for some or all of the electricity grids and the electric vehicles for any given time point or interval.”).
Vinson, Bulut, Wu, Conley, and Bhimani are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu, further in view of Conley to incorporate the teachings of Bhimani and utilize carbon intensity data as a data feature. A person of ordinary skill in the art would recognize that data centers are a significant source of carbon output due to massive energy consumption, and would be motivated by the need to minimize the carbon footprint of data centers.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Ghiasi et al. (US 20060253715 A1) hereafter Ghiasi.
Regarding claim 6, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 2.
Wu teaches:
a plurality of servers (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”);
the sleep state solution (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Vinson in view of Bulut further in view of Wu further in view of Conley does not teach includes a voltage and scaling mode for each server, wherein the voltage and scaling mode are associated with compute ability and power consumption.
However, Ghiasi teaches:
includes a voltage and scaling mode for each server, wherein the voltage and scaling mode are associated with compute ability and power consumption (Paragraph 9; “The solution explores the use of dynamic voltage scaling to respond to changes in server demands”. Paragraph 31 further discloses “FIG. 4 illustrates the results of the operation of the invention on an example cluster data processing system in which work is assigned to a plurality of nodes and the processing speed and power consumption of the nodes are controlled in accordance with an exemplary embodiment of the present invention”, where the use of dynamic voltage scaling corresponds to the voltage and scaling mode and the assignment of the results based on processing speed and power consumption correspond to association with compute ability and power consumption).
Vinson, Bulut, Wu, Conley, and Ghiasi are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley to incorporate the teachings of Ghiasi and have the sleep state solution include a voltage and scaling mode for each server in the plurality of servers and have them associated with compute ability and power consumption. A person of ordinary skill in the art would have recognized known concept of voltage and scaling in server management and the implementation of which would predictably yield optimized server deployment configuration.
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Dey et al. (US 20190363988 A1) hereafter Dey.
Regarding claim 10, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 9. Vinson in view of Bulut further in view of Wu further in view of Conley does not teach wherein the server availability schedule is implemented by a load balancer to a subset of servers of the plurality of servers of the computing system.
However, Dey teaches:
wherein the server availability schedule is implemented by a load balancer to a subset of servers of the plurality of servers of the computing system (Paragraph 40; “the gateway load balancer 304 checks the health of each server instance before assigning a request. In this way, certain example embodiments may narrow the pool of servers or server instances to which requests can be distributed, in essence applying the conventional algorithm to a partial subset of the server instances in the overall cluster” directly corresponds to a load balancer applying the schedule to a subset of servers).
Vinson, Bulut, Wu, Conley, and Dey are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley to incorporate the teachings of Dey and have the server availability schedule implemented by a load balancer to a subset of servers of the system. A person of ordinary skill in the art would recognize that not all servers would need to receive the new schedule, and therefore it would be a logical design decision to have applied the updated schedule to only a subset thereof.
Claims 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Baskaran et al. (US 11715051 B1) hereafter Baskaran.
Regarding claim 11, Vinson in view of Bulut, further in view of Wu, further in view of Conley teach the method of claim 9. Wu teaches:
the sleep state solution (Paragraph 39; “server state manager 620 can receive the output from the decision engine 615 and in response can adjust the current capacity to meet the required capacity by turning down active servers if the current capacity is greater than the required capacity or turning up additional servers if the current capacity is lower than the required capacity”, where the operational state assignment for each server in the server pool corresponds to the sleep state solution).
Conley teaches:
the solutions comprise a plurality of recommendations (Col. 11, line 59 – Col. 12, line 10; “user interface may present one or more notifications or recommendations”, “a recommendation may indicate that a particular electrical device 102 is not normally used between the hours of 12:00 P.M. and 12:00 A.M. and suggest that the electrical device 102 be transitioned to a low power state”, where “[t]he notification may indicate a quantity of electrical power or a financial cost that may be conserved by placing particular electrical devices 102 in a low power state during particular time periods.”, the one or more recommendations thereby contemplating a plurality of recommendations.).
Vinson teaches:
analyzing the plurality of sleep state recommendations of the optimized sleep state solution (Paragraph 86; “the outputs 730 include a schedule recommendation 732 and a confidence score 734” shows that the scheduling recommendations are outputs of the ML scheduling service and subject to analysis, corresponding to analyzing recommendations);
determining at least one sleep state recommendation of the plurality of sleep recommendations is not used when implementing the optimized sleep state solution (Paragraph 89; “the ML scheduling service can decline to replace an existing schedule… with an updated schedule if the confidence score is below a threshold”, where by declining to apply certain recommendations (low-confidence ones), the system determines that some recommendations are not used.);
and refactoring the optimized sleep state solution by removing the at least one sleep state recommendation (Paragraph 89; “the ML scheduling service can decline to replace an existing schedule… with an updated schedule if the confidence score is below a threshold”, where declining to adopt certain schedules effectively removes them from the optimized solution, corresponding to refactoring the optimized sleep state solution.).
Vinson in view of Bulut further in view of Wu further in view of Conley does not explicitly teach underutilization.
However, Baskaran teaches:
underutilization (Col. 188, lines 38-44; “In the example provided in FIG. 52B, the cluster 5253 values are shown as “0”, “1”, or “2”. A value of “0” may correspond to a severely underutilized (e.g., shutdown) recommendation, a value of “1” may correspond to an underutilized (e.g., downgrade) recommendation, and a value of “2” may corresponding an over utilized (e.g., upgrade) recommendation”).
Vinson, Bulut, Wu, Conley, and Baskaran are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley to incorporate the teachings of Baskaran and have the refactoring occur when underutilization is determined. A person of ordinary skill in the art would recognize the need to retrieve memory space utilized to store possible recommendations as a known method in the art yielding the predictable result of preventing an out of memory error occurring from too many recommendations being stored.
Regarding claim 12, Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Baskaran teach the method of claim 11. Vinson teaches:
wherein the refactoring is initiated based on one or more rules being met (Paragraph 89; “These thresholds can be determined manually (e.g., by an administrator) or automatically (e.g., using the scheduling ML model 124 or another suitable ML model or software techniques)”, where the thresholds act as rules that control whether recommendations are kept or discarded, corresponding to initiating refactoring based on one or more rules being met.).
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Baskaran, further in view of Kostyuk et al. (US 20230401482 A1) hereafter Kostyuk.
Regarding claim 13, Vinson in view of Bulut, further in view of Wu, further in view of Conley, further in view of Baskaran teach the method of claim 12. Vinson teaches:
wherein a first rule initiates refactoring if a given sleep state recommendation is not selected over a time period (Paragraph 89; “the ML scheduling service can decline to replace an existing schedule… with an updated schedule if the confidence score is below a threshold”, where a confidence score threshold effectively functions like a temporal/usage rule. If a schedule is not selected, it triggers removal/refactoring).
Vinson in view of Bulut further in view of Wu further in view of Conley further in view of Baskaran does not teach an explicit time period.
However, Kostyuk teaches:
a time period (Paragraph 72; “The computing machine may receive data every threshold time period (e.g., daily) and may use that data to train the agents to continually improve recommendations based on observed results”).
Vinson, Bulut, Wu, Conley, Baskaran, and Kostyuk are considered to be analogous to the claimed invention because they are in the same field of resource management. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Vinson in view of Bulut further in view of Wu further in view of Conley further in view of Baskaran to incorporate the teachings of Kostyuk and have the refactoring occur based on a time period. A person of ordinary skill in the art would recognize the use of a time period to determine whether a recommendation should be dropped to be a known method in the art yielding the predictable result of freeing memory in use by recommendations and to avoid out of memory errors occurring from too many recommendations being stored.
Response to Arguments
Applicant's arguments filed 12/10/2025 have been fully considered. Applicant’s arguments are summarized below:
The prior art references of record do not disclose, teach, or suggest the amended independent claim features of “historical data features to determine sleep state solutions for each server in the plurality of servers based on a transition delay and power characteristics for each server in the plurality of servers”.
Dependent claims are submitted as allowable for at least the above reasons.
Examiner’s response:
The examiner agrees that the previous prior art references of record do not disclose the amended independent claim 1 features due to the sleep states of Tsirkin being based on processor sleep states and not applicable to server sleep states. Therefore, the previous rejections under 35 U.S.C. 103 are withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Vinson, Bulut, Wu, and Conley under 35 U.S.C. 103.
Independent claims 1, 14, and 18 remain rejected for the reasons stated above. Therefore, contrary to Applicant's arguments, because the dependent claims depend from an unpatentable claim and does not add limitations that overcome the rejection, it likewise remains rejected.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Rabelo et al. (NPL: Development of a Real-Time Learning Scheduler Using Reinforcement Learning Concepts; August 1994) discusses the use of reinforcement learning algorithms to perform resource scheduling in dynamic environments.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action. Any inquiry concerning this communication or earlier communications from the examiner should be directed to KENNETH P TRAN whose telephone number is (571)272-6926. The examiner can normally be reached M-TH 4:30 a.m. - 12:30 p.m. PT, F 4:30 a.m. - 8:30 a.m. PT, or at Kenneth.Tran@uspto.gov.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, April Blair can be reached at (571) 270-1014. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/KENNETH P TRAN/ Examiner, Art Unit 2196
/APRIL Y BLAIR/ Supervisory Patent Examiner, Art Unit 2196