International Business Machines Corporation

Yaacov Frank

Mudi M Fluman

Yehuda Shiran

Ronny Vatelmacher

Igor Nabutovsky

Janice Marie Girouard

Shay Saffer

Rivka Mayraz Matosevich

Osnat Shasha

Zah Barzik

Amit Margalit

Ofir Elizov

Tomer Goldberg

A technique for using machine logic to configure power supply units in a data center including: (i) receiving cost information for multiple power grids that includes expected cost information for a time period; (ii) receiving usage information for multiple data center loads in a data center that includes expected electrical power usage for the data center loads; (iii) determining a target power information set, based, at least in part, on the cost information and the usage information, with the target power information set, for the time period, including a target power amount respectively corresponding to each power grid; and (iv) determining a power supply unit configuration scheme so that, when operating under the power supply unit configuration scheme, each power grid will supply, at least approximately, its respectively corresponding target power amount.

Intermittently redistributing energy from multiple power grids in a data center context

Intermittently re-distributing power from three phases of a power grid connected to power consuming components of a data center by configuring power supply units (PSUs) which are integrated into each enclosure, two PSUs per enclosure. The distribution is determined by machine logic to: (i) meet the predicted power requirements of the power consuming components of the data center, and (ii) the use of first phase, second phase and third phase electrical powers is at least approximately in balance.

Continuously balancing multiple phases of a power grid

A computer receives data from a storage device. The computer identifies a state value of the storage device and at least one maintenance task included in the received data. The computer generates a set of instructions for limiting the workload of a storage device executing maintenance tasks. According to the generated instructions, the computer limits the workload placed on the storage device.

Communicating outstanding maintenance tasks to improve disk data integrity

Methods, computing systems and computer program products implement embodiments of the present invention that include determining a required electrical power level for one or more computer elements. Upon selecting a first value indicating a first electrical power level that is less than the required electrical power level, the first value is conveyed to a first power supply unit (PSU), so that the first PSU delivers, to the one or more computer elements, the first electrical power level. A second value is selected that indicates a second electrical power level that is different from the first value, the second value indicating a difference between the required electrical power level and the first electrical power level. The second value is conveyed to a second PSU, so that upon receiving the second value, the second PSU delivers, to the one or more computer elements, the second electrical power level.

Unbalanced load sharing with parallel power supplies

Storage operation requests from any device of a computing environment can be numerous and frequent. In particular, if there is a high frequency initiation of storage operation requests to store, retrieve, or modify data, then targeted storage systems have to easily and quickly decide in which order to satisfy the storage operation requests, such as when two requests occur to retrieve identical data. Storage operation requests can be prioritized at the end device instead of any intermediary device or enabling a complex ordering algorithm. Moreover, the storage on a cloud model consists of similar storage services which serve consumers of different needs. Some applications/users can afford longer service time than other applications/users. Differentiation in required service time allows price differentiation. The solution will serve premium customers faster than it serves customers who paid less.

Prioritizing storage operation requests utilizing data attributes

In response to a warning that power may be interrupted, a non-volatile data storage sub-system of a host computer system re-orders machine readable instructions that the non-volatile data storage sub-system is going to perform. This re-ordering of instructions decreases the probability that important data will be lost. The re-ordering of instructions is performed according to rules.

Data storage device control with power hazard mode

A computer determines an intrinsic read speed and an intrinsic write speed associated with a first disk and a second disk. The computer receives a request to read a portion of data, wherein the portion of data is stored redundantly on both the first and second disk. The computer identifies a first latency associated with reading the portion of data from the first disk, where the first latency is based on at least the intrinsic read speed and the intrinsic write speed associated with the first disk. The computer identifies a second latency associated with reading the portion of data from the second disk, wherein the second latency is based on at least the intrinsic read speed and the intrinsic write speed associated with the second disk. The computer determines that the first latency exceeds the second latency. The computer selects the second disk to read the portion of data.

Performance-based grouping of storage devices in a storage system

Embodiments of the present invention provide systems, methods, and computer program products for implementing multiple raid level configurations in a computer storage device. In one embodiment, performance or resiliency of application data being executed to a single computer storage device can be prioritized. Embodiment of the present invention provide systems, methods, and computer program products for a recovery operation, responsive to determining to prioritize performance of application data being executed to the single computer storage device.

Implementing multiple raid level configurations in a data storage device

Methods, computing systems and computer program products implement embodiments of the present invention that include collecting, from a software-defined storage system including one or more computing nodes that are configured to provide a storage service, performance metrics for each of the computing nodes, and detecting, based on the performance metrics, a performance bottleneck in the software-defined storage system. In embodiments of the present invention, each of the computing nodes includes one or more software modules and one or more hardware components, and the performance bottleneck is either a given software module or a given hardware component. In some embodiments, detecting the performance bottleneck includes predicting the performance bottleneck. Upon detecting the performance bottleneck, a solution for the performance bottleneck can be determined, and the performance bottleneck and the solution can be presented to a user on a display.

Software-defined storage system monitoring tool

Methods, computing systems and computer program products implement embodiments of the present invention that include determining a required electrical power level for one or more computer elements. Upon selecting a first value indicating a first electrical power level that is less than the required electrical power level, the first value is conveyed to a first power supply unit (PSU), so that that the first PSU delivers, to the one or more computer elements, the first electrical power level. A second value is selected that indicates a second electrical power level that is different from the first value, the second value indicating a difference between the required electrical power level and the first electrical power level. The second value is conveyed to a second PSU, so that that upon receiving the second value, the second PSU delivers, to the one or more computer elements, the second electrical power level.

Embodiments of the present invention provide an apparatus and methods for converting a volatile memory module device to a non-volatile memory module device. In one embodiment, a male contact edge of a memory module device is inserted to a first memory module device socket of a memory module device adaptor, the memory module device adaptor having a first male contact edge. Power is provided to the memory module device in the event of a system failure. Diodes operatively connected to one or more power pins of the first memory module device socket prevent current from flowing to the battery to one or more power pins of the male contact edge of the memory module device adaptor. Applications needing dedicated memory are registered and, in the event of a system failure memory module device data associated with the registered applications can be recovered without the use of flash memory.

Converting volatile memory module devices to flashless non-volatile memory module devices

A computer determines an intrinsic read speed and an intrinsic write speed associated with a first disk and a second disk. The computer receives a request to read a portion of data, wherein the portion of data is stored redundantly on both the first and second disk. The computer identifies a first latency associated with reading the portion of data from the first disk, where the first latency is based on at least the intrinsic read speed and the intrinsic write speed associated with the first disk. The computer identifies a second latency associated with reading the portion of data form the second disk, wherein the second latency is based on at least the intrinsic read speed and the intrinsic write speed associated with the second disk. The computer determines that the first latency exceeds the second latency. The computer selects the second disk to read the portion of data.

Growing community of inventors

Ramot Meir, Israel

Yaacov Frank