International Business Machines Corporation

Lenovo (singapore) Pte. Ltd.

Carnegie Mellon University

Rajarshi Das

Jeffrey Owen Kephart

Canturk Isci

Jonathan Lenchner

Hendrik F Hamann

Hoi Yeung Chan

Philip Lee Vitale

James Edwin Hanson

Gerald James Tesauro

David W Levine

Bo Yang

Murthy V Devarakonda

Omar Cardona

Ian N Whalley

Steve Richard White

Alla Segal

Nikhil Hegde

Michael Elton Nidd

Aaron Christoph Sawdey

Bret W Lehman

Christopher L Molloy

Suzanne K McIntosh

Metin Feridun

Axel Tanner

Vanessa Lopez-Marrero

Andriy Stepanchuk

Srinivas Yarlanki

Tony A Bentivegna

Robert G Farrell

Nitendra Rajput

Juan C Rubio

Marc Adam Kaplan

Wael R El-Essawy

Thomas W Keller, Jr

Yixin Diao

Brad L Brech

Saurabh Bagchi

Vishal Ramachandra Mansur

Kilian Quirin Weinberger

Anshul Gandhi

Mor Harchol-Balter

Edwin Richie Lassettre

Vishal R Mansur

Bradley L Brech

William Edward Walsh

Nicholas K Jong

Techniques for providing an evaluation for an evaluation subject are described. The techniques include verifying the accuracy of evaluations based on an audio or recorded evaluation. The evaluations are verified by identifying a set of evaluators to provide an evaluation of an evaluation subject and requesting the evaluation from one of the identified set of evaluators. The evaluators then provide to a system described herein a structured evaluation input and an audio evaluation input from the solicited evaluator. The audio evaluation input is processed for evaluation indicators and the structured evaluation input is verified based on the evaluation indicators.

Performance evaluation using audio and structured feedback

Apparatuses, methods, systems, and computer program products are disclosed for monitoring performance of multithreaded workloads. A workload module starts a workload comprising one or more threads. The workload reaches a steady state after a period of time. A data module receives a first set of performance data for each thread in response to the workload reaching the steady state. The first set of performance data is received concurrently for each thread. The data module receives a second set of performance data for each thread a period of time after the first set of performance data is received. The second set of performance data is received concurrently for each thread. A result module calculates one or more performance metrics for each thread based on the first and second sets of performance data.

Monitoring performance of multithreaded workloads

Apparatuses, methods, systems, and computer program products are disclosed for calculating a clock rate of a processor. A baseline data module receives a first set of performance data associated with a processor. The performance data is generated using a hardware element that captures performance data for the processor. The hardware element is external to the processor. An update data module receives a second set of performance data associated with the processor a predefined time interval after the first set of performance data is received. The second set of performance data corresponds to the first set of performance data. A rate module calculates a clock rate for the processor based on the first set of performance data and the second set of performance data.

Calculating the clock frequency of a processor

A method for storing service level agreement ('SLA') compliance data includes reserving a memory location to store SLA compliance data of a software thread. The method includes directing the software thread to run on a selected hardware device. The method includes enabling SLA compliance data to be stored in the memory location. The SLA compliance data is from a hardware counting device in communication with the selected hardware device. The SLA compliance data corresponds to operation of the software thread on the selected hardware device.

Storing service level agreement compliance data

An apparatus for storing service level agreement ('SLA') compliance data is disclosed. A method and a computer program product also perform the functions of the apparatus. The apparatus includes a reservation module that reserves a memory location to store SLA compliance data of a software thread. The apparatus includes a directing module that directs the software thread to run on a selected hardware device. The apparatus includes an enabling module that enables SLA compliance data to be stored in the memory location. The SLA compliance data is from a hardware counting device in communication with the selected hardware device. The SLA compliance data corresponds to operation of the software thread on the selected hardware device. At least a portion of the reservation, the module, and the enabling modules includes one or more of hardware and program instructions. The program instructions are stored on one or more computer readable storage media.

According to one aspect of the present disclosure, a method and technique for dynamic reconfiguration of network devices for outage prediction is disclosed. The method includes: aggregating a plurality of network devices to form an input/output (I/O) link bundle; identifying a quality of service level of data throughput needed by at least one workload utilizing the I/O link bundle; monitoring a parameter of each of the network devices of the I/O link bundle; determining whether the monitored parameter indicates a predicted data throughout quality of service level decrease by a network device being utilized by the workload below the quality of service level for the workload; and responsive to determining that the monitored parameter indicates a predicted quality of service level decrease by the respective network device, dynamically reconfiguring the link bundle to maintain the quality of service level for the workload utilizing the I/O link bundle.

Dynamic reconfiguration of network devices for outage prediction

Methods and apparatus are provided for automated object classification using temperature profiles. An object in an environment (such as an exemplary data center) is classified by obtaining a surface temperature profile of the object; and classifying the object as a particular type of equipment based on the obtained surface temperature profile. The surface temperature profile of the object can be compared to a plurality of predefined characteristic surface temperature profiles each associated with a given type of equipment.

Automated object classification using temperature profiles

Methods, apparatuses, and computer program products for provisioning aggregate computational workloads and air conditioning unit configurations to optimize utility of air conditioning units and processing resources within a data center are provided. Embodiments include for each air conditioning unit within the data center, determining a thermal zone generated by the air conditioning unit; for a given aggregate computational workload, identifying a plurality of computational workload configurations, each computational workload configuration indicating spatial assignments of the aggregate computational workload among a plurality of processing resources within the data center; for each computational workload configuration, calculating a total minimum energy consumption of the air conditioning units and the processing resources; and selecting the computational workload configuration with the lowest total minimum energy consumption and the determined lowest power air conditioning unit configuration corresponding with the selected computational workload configuration.

Provisioning aggregate computational workloads and air conditioning unit configurations to optimize utility of air conditioning units and processing resources within a data center

Techniques are described for controlling the climate in a data center. Using the input of an administrator, multiple desired attributes of a data center (e.g., temperature, energy consumption, costs, or system performance) may be balanced using a utility function that maximizes the utility of the computing systems in the data center according to the administrator's preferences. Additionally, a cooling model is generated that represents the affect of a control parameter (e.g., the fan speed of a CRAC) on the desire attributes of the data center. The cooling model may then be substituted into the utility function to replace the desired attributes. Using this new utility function, the control parameters may be varied such that the maximum utility is achieved.

Unified and flexible control of multiple data center cooling mechanisms

A system and technique for dynamic reconfiguration of network devices for outage prediction is disclosed. The system includes a host and a plurality of network devices aggregated to form a link bundle; and wherein the host comprises a configuration manager operable to: identify a quality of service level of data throughput needed by at least one workload utilizing the link bundle; monitor a parameter of each of the network devices of the link bundle; determine whether the monitored parameter indicates a predicted data throughout quality of service level decrease by a network device being utilized by the workload below the quality of service level for the workload; and responsive to determining that the monitored parameter indicates a predicted quality of service level decrease by the respective network device, dynamically reconfigure the link bundle to maintain the quality of service level for the workload utilizing the link bundle.

Mechanisms are provided for determining the location of computing assets within an organization. These mechanisms determine first location information identifying a location of at least one first physical asset housing in a predetermined physical area of the organization relative to a predetermined coordinate system, and second location information identifying a location of at least one second physical asset housing within the at least one first asset housing. The mechanisms automatically determine third location information identifying a location of at least one computing asset within the at least one second physical asset housing. In addition, the mechanisms generate an asset location map data structure for the organization based on the first, second and third location information. The mechanisms also perform at least one management operation for managing resources of the organization, based on the asset location map data structure for the organization.

Location of computing assets within an organization

Growing community of inventors

Armonk, NY, United States of America

Rajarshi Das