International Business Machines Corporation

Brian Chase Twichell

Sergio Reyes

Samuel K Ingram

Yijie Zhang

Michael Jay Shapiro

Brent William Yardley

Faisal Ahmed

Diane Garza Flemming

David Blair Whitworth

Greg Robert Mewhinney

Frank Eliot Levine

Bruce Gerard Mealey

Geraint M North

Edward Hugh Welbon

Adam James McNeeney

Ann Marie Maynard

Computer-implemented methods for optimized compute resource addition and removal in a distributed storage platform. In a case of a newly added compute resource being connected to a storage subsystem shared by compute resources in the distributed storage platform, the distributed storage platform formulates a redistribution plan to redistribute a subset of a global address space of the storage subsystem to a newly added logical volume in the storage subsystem. In a case of a removed compute resource being disconnected from the storage subsystem, the distributed storage platform formulates a redistribution plan to redistribute respective logical blocks in a logical volume for the removed compute resource to respective remaining logical volumes for respective remaining compute resources in the distributed storage platform. The distributed storage platform executes the redistribution plan to reassign data block ownerships on one or more physical memory devices in the storage subsystem.

Optimized addition and removal of compute resources in a distributed storage platform by implementing mapping changes in a shared storage subsystem

A mechanism is provided for programmatic implicit multithreading. A first operation is executed on a first thread in a processor, where the first operation is from a set of operations within a block of code of an application that are distinct and process unrelated data. A determination is made as to whether a time limit associated with executing the first operation has been exceeded. Responsive to the time limit being exceeded, a determination is made as to whether there is one or more unexecuted operations in the set of operations. Responsive to one or more unexecuted operations existing in the set of operations, a new thread is spawned off on the processor to execute a next unexecuted operation of the one or more unexecuted operations.

Programmatic implicit multithreading

Providing access to a data storage resource. A storage subsystem comprising one or more storage address units and is associated with one or more access interfaces is identified. An address-interface correlation guideline is identified that defines a combination of rules that govern which access interfaces are used to access storage address units. A target address unit identification is received from a requesting system. A processor determines which storage address units a requesting system requests to access to based on the received target address unit identification. The target address unit identification is associated with at least one of the storage address units. The requesting system is provided with access to the storage address units using access interfaces that are determined based on a target interface conclusion.

Managing program access to data storage resources

A data storage system includes first and second storage devices accessed via first and second hardware channels, respectively. A storage controller receives a first access request requesting access to first data and a second access request requesting access to second data, where both the first data and the second data are accessed via the first hardware channel. In response to detecting receipt of the first and second access requests within a same service window defining a range of temporal proximity, the storage controller records an association of identifiers of the first data and second data requested by the first and second access requests, respectively. The storage controller thereafter migrates the second data accessed by the second access request from the first storage device to the second storage device based on the recorded association, such that the first and second data are available for access via different hardware channels.

Data storage system performing data relocation based on temporal proximity of accesses

An apparatus includes a particle trap coupled to a first surface of an enclosure, wherein the first surface of the enclosure is opposite a top surface of a circuit board. A particle guard coupled to the top surface on a first side of the circuit board located in the enclosure, wherein the enclosure includes one or more apertures on a second surface of the enclosure where the first side of the circuit board is introduced to an external airflow.

Dust guard structure

An amplification number may be input into a storage subsystem interface. A processor in a storage system may receive an original input/output (I/O) request from an application. The processor may determine, in response to the amplification number being input, to duplicate the original I/O request one or more times. The processor may generate one or more duplicate I/O requests of the original I/O request. The processor may store the original I/O request in a general address space in the storage subsystem. The processor may store the one or more duplicate I/O requests in a reserved address space in the storage subsystem. The processor may execute the original I/O request and the one or more duplicate requests.

I/O amplification for determining to increase workload

Storage system and application intercommunication is provided. A first kernel-mode module determines a first event corresponding to an operational parameter of a first node based, at least in part, on a shared namespace accessible by the first kernel-mode module and a second kernel-mode module. An interrupt is issued based on the first event from the first kernel-mode module to the second kernel-mode module via an interface. The second kernel-mode module issues a second event to a second node, wherein the second event corresponds to the object of the shared namespace.

Storage and application intercommunication using ACPI

An apparatus for a dust guard structure includes a particle guard coupled to a top surface of a circuit board, wherein the particle guard is located along a first side of the circuit board between an edge of the circuit board and a first electronic component. The dust guard structure further includes the first electronic component electrically coupled to the circuit board via one or more electronic connections, wherein a height of the particle guard is greater than a height of each of the one or more electrical connections of the first electronic component. The dust guard structure further includes the first side of the circuit board being introduced to an external airflow.

A computer-implemented method comprises identifying a storage subsystem comprising one or more storage address units and associated with one or more access interfaces; identifying an address-interface correlation conclusion; and identifying a target address unit identification associated with at least of the one or more storage address units. The computer-implemented method further comprises determining a target interface conclusion associated with at least one of the one or more access interfaces based on the address-interface correlation guideline and the target address unit identification. A corresponding computer program product and computer system are also disclosed.

Determining a preferred interface for write access to a data storage system having multiple interfaces. Interface preference is determined at the data-stripe level. Write requests are routed to the preferred interface.

Raid-topology-aware multipath routing

A method, system and program are provided for controlling access to a specified cache level in a cache hierarchy in a multiprocessor system by evaluating cache statistics for a specified application at the specified cache level against predefined criteria to prevent the specified application from accessing the specified cache level if the specified application does not meeting the predefined criteria.

Use of cache statistics to ration cache hierarchy access

Performing storage writes in a mirrored virtual machine system by receiving a state of a primary virtual machine during execution of an application, wherein the primary virtual machine runs on a first physical machine and a secondary virtual machine runs on a second physical machine, wherein the state is captured by checkpointing, and the primary virtual machine is configured to write data to a first block and concurrently write the data to a write buffer on the secondary virtual machine. The method also includes storing a copy of data within a second block to a rollback buffer for the secondary virtual machine, in response to identifying a checkpoint in the application, merging the rollback buffer with the write buffer, in response to detecting a failover, writing a copy of the rollback buffer to the disk storage, and continuing execution on the secondary virtual machine from the last checkpoint.

Growing community of inventors

Austin, TX, United States of America

Brian Chase Twichell