Abstract: A system for generating a power consumption model of at least one server includes one or more computers configured to obtain n time series telemetry signals indicative of operating parameters of the at least one server, obtain a time series power signal indicative of power consumed by the at least one server, and correlate each of the n time series telemetry signals with the time series power signal. The one or more computers are further configured to select a set of the n time series telemetry signals having an overall correlation with the time series power signal greater than a predetermined threshold, and generate a power consumption model of the at least one server based on at least the set of the n time series telemetry signals.

Abstract: A system for controlling energy usage in a server having a processor, where the system includes a memory for storing energy cost information, and a controller for determining a transaction rate for the processor. The controller is also for determining a cumulative of energy expended by the server based on the determined transaction rate for each of a number of available power level states (P-states) for operation of the processor, and for selecting one of the available P-states for operation of the processor based on the determined cumulative energy expended and the stored energy cost information.

Abstract: The disclosed embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data, which includes first information containing telemetric signals gathered using sensors in the computer system and second information that indicates one or more transaction latencies of software running on the computer system. Upon detecting an upward trend in the one or more transaction latencies, the system analyzes the telemetry data for a correlation between the one or more transaction latencies and one or more environmental factors represented by a subset of the telemetric signals. Upon identifying the correlation between the one or more transaction latencies and an environmental factor, the system stores an indication that the environmental factor may be contributing to the upward trend in the one or more transaction latencies.

Abstract: A system for controlling energy usage in a server having a processor, where the system includes a memory for storing energy cost information, and a controller for determining a transaction rate for the processor. The controller is also for determining a cumulative of energy expended by the server based on the determined transaction rate for each of a number of available power level states (P-states) for operation of the processor, and for selecting one of the available P-states for operation of the processor based on the determined cumulative energy expended and the stored energy cost information.

Abstract: Computer system fans having fixed operating states corresponding to discreet operating speeds may be controlled by collecting temperature information upstream or downstream of the fans and commanding the fans to switch between the fixed operating states based on the temperature information at a frequency sufficient to controllably achieve speeds between the discreet operating speeds.

Abstract: Some embodiments of the present invention provide a system that determines whether a cooling device in a computer system is responsive to control signals. During operation of the computer system, a control signal is sent to the cooling device. Next, a response of the computer system to the control signal is measured, wherein the response includes a temperature profile. The frequency content of the control signal is then compared to the frequency content of the temperature profile to determine whether the cooling device is responsive to the control signal.

Abstract: The disclosed embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data as a set of telemetric signals using a set of sensors in the computer system. Next, the system analyzes the telemetry data to estimate a value of a parameter associated with the computer system, wherein the parameter is at least one of a power utilization and a temperature. Finally, the system controls a subsequent value of the parameter by modulating a virtual duty cycle of a processor in the computer system based on the estimated value.

Abstract: A method for identifying missing components of a computer system may include receiving telemetry signals characterizing a current configuration of the computer system and determining a cross power spectral density signature of at least some of the telemetry signals. The method may further include comparing information about the determined cross power spectral density signature with information about a predetermined cross power spectral density signature to determine whether a component is missing within the computer system.

Abstract: Some embodiments of the present invention provide a system that controls a device that generates vibrations in a computer system. During operation, a critical vibration frequency is determined for the computer system. Next, a keep-out zone is generated based on the critical vibration frequency, wherein the keep-out zone specifies a range of frequencies to be avoided. Then, the device is controlled based on the keep-out zone to reduce vibrations generated by the device in the keep-out zone.

Abstract: Some embodiments of the present invention provide a system that controls a cooling fan for a storage array. During operation, an input-output (I/O) metric of the storage array is monitored. Then, the cooling fan is controlled based on the I/O metric.

Abstract: Leaked memory in a computer system is detected and recovered by first detecting memory leakage within the computer system based on nonlinear and non-parametric time-series regression analysis of software telemetry data generated by one or more software process running on the computer system. If existence of memory leakage is detected, then memory that has leaked is specifically identified and recovered. This is done by halting one or more of the software processes, generating a core image file or files of the halted software process(es), and re-starting the halted process or processes without waiting for analysis of the core image file(s). Then, the core image file is evaluated to specifically identify leaked memory in the computer system based on the core image file. Finally, the identified leaked memory is recovered.

Abstract: The disclosed embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data as a set of telemetric signals using a set of sensors in the computer system. Next, the system uses a regularization technique to calculate a temperature derivative with respect to time for a component in the computer system from the telemetric signals. Finally, the system controls a subsequent value of the temperature derivative with respect to time by modulating a fan speed in the computer system based on the calculated temperature derivative with respect to time and the telemetric signals.

Abstract: Some embodiments of the present invention provide a system that accurately synchronizes signals related to the operation of a computer system. During operation, the system receives a first time-domain signal associated with a first system variable and a second time-domain signal associated with a second system variable from the computer system. The system then transforms the first and the second time-domain signals into a first frequency-domain signal and a second frequency-domain signal, respectively. Next, the system computes a cross-power-spectral-density (CPSD) between the first and second frequency-domain signals to obtain a phase angle versus frequency graph between the two frequency-domain signals. The system subsequently extracts the slope of the phase angle versus frequency graph, and uses the value of the slope to synchronize the first time-domain signal and the second time-domain signal.

Abstract: Some embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data as a set of telemetric signals using a set of sensors in the computer system, wherein the set of sensors includes temperature sensors at different locations in a cooling airflow through the computer system. Next, the system calculates a set of pairwise temperature differences from the telemetric signals. The system then infers an altitude of the computer system based on the pairwise temperature differences. Finally, the system uses the inferred altitude to manage the operation of the computer system.

Abstract: One embodiment of the present invention provides a system that trains a pattern-recognition model for electronic prognostication for a computer system. First, the system monitors a performance parameter from a set of computer systems that includes at least two computer systems, wherein monitoring the performance parameter includes systematically monitoring and recording performance parameters in a set of performance parameters from computer systems in the set of computer systems, wherein the recording process keeps track of the temporal relationships between events in different performance parameters in the set of performance parameters. Next, the system generates a training data set based on the monitored performance parameter from the set of computer systems, wherein generating the training data set includes concatenating two or more time-series of the performance parameter from computer systems in the set of computer systems.

Abstract: A system for determining vibration of at least one hard disk drive includes one or more computers configured to command a plurality of reads from the at least one hard disk drive such that, for each of the plurality of reads, data stored on the at least one hard disk drive is retrieved directly from the at least one hard disk drive. The one or more computers are also configured to obtain read rate information related to the plurality of reads and to determine a vibration level experienced by the at least one hard disk drive based on the read rate information.

Abstract: Some embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data as a set of telemetric signals from the computer system and validates the telemetric signals using a nonlinear, nonparametric regression technique. Next, the system assesses the integrity of a power supply unit (PSU) in the computer system by comparing the telemetric signals to one or more reference telemetric signals associated with the computer system. If the assessed integrity falls below a threshold, the system performs a remedial action for the computer system.

Abstract: The disclosed embodiments provide a system that analyzes telemetry data from a computer system. During operation, the system obtains the telemetry data as a set of telemetric signals using a set of sensors in the computer system. Next, the system analyzes the telemetry data to estimate a value of a parameter associated with the computer system, wherein the parameter is at least one of a power utilization and a temperature. Finally, the system controls a subsequent value of the parameter by modulating a virtual duty cycle of a processor in the computer system based on the estimated value.

Abstract: Some embodiments of the present invention provide a system that synchronizes signals related to the operation of a computer system. During operation, a set of correlation coefficients between a first signal and a second signal is generated, wherein each correlation coefficient is associated with a different phase shift between the first signal and the second signal. Then, a synchronizing phase shift associated with the highest correlation coefficient in the set of correlation coefficients is determined in order to synchronize the first signal and the second signal.

Abstract: A method for identifying missing components of a computer system may include receiving telemetry signals characterizing a current configuration of the computer system and determining a cross power spectral density signature of at least some of the telemetry signals. The method may further include comparing information about the determined cross power spectral density signature with information about a predetermined cross power spectral density signature to determine whether a component is missing within the computer system.