Hewlett Packard Enterprise Development LP

Reldata, Inc.

Silicon Graphics International Corporation

Quantaro, LLC

Kirill Malkin

Michael S Woodacre

Steven J Dean

Eric Carl Fromm

Randal Steven Passint

Vladislav Bolkhovitin

Thomas Edward McGee

Eng Lim Goh

Michael E Malewicki

Steve Dean

Oleg Neverovitch

Matthew Jacob

Mikhail Litvin

Alan Poston

An ultraconverged architecture has multiple availability zones within a single server. The functionality in each of the availability zones is independently controlled, such that resetting and/or disconnecting any component in any availability zone from power and replacing said component does not affect availability of any other component in any other availability zone. A manager of availability zones controls reset functionality in each of a plurality of availability zones. The manager of availability zones generates a requested reset type in the requested availability zone. The manager of availability zones generates reset signals or requests for some or all components located in multiple availability zones. The reset signal or request is generated upon external request to the manager of availability zones that specifies the reset type, the availability zone, and optionally the list of components to be reset. The manager of availability zones discovers and enumerates the components in each availability zone.

Ultraconverged systems having multiple availability zones

A scalable software stack is disclosed. In particular, the present disclosure provides a system and a method directed at allocating logical ownership of memory locations in a shared storage device among two or more associated compute devices that have access to the storage device. The logical ownership allocation can minimize potential conflicts between two simultaneous accesses occurring within the same memory location of the storage device.

Scalable software stack

In high performance computing, the potential compute power in a data center will scale to and beyond a billion-billion calculations per second ('Exascale' computing levels). Limitations caused by hierarchical memory architectures where data is temporarily stored in slower or less available memories will increasingly limit high performance computing systems from approaching their maximum potential processing capabilities. Furthermore, time spent and power consumed copying data into and out of a slower tier memory will increase costs associated with high performance computing at an accelerating rate. New technologies, such as the novel Zero Copy Architecture disclosed herein, where each compute node writes locally for performance, yet can quickly access data globally with low latency will be required. The result is the ability to perform burst buffer operations and in situ analytics, visualization and computational steering without the need for a data copy or movement.

Method and system for shared direct access storage

A high performance computing (HPC) system has an architecture that separates data paths used by compute nodes exchanging computational data from the data paths used by compute nodes to obtain computational work units and save completed computations. The system enables an improved method of saving checkpoint data, and an improved method of using an analysis of the saved data to assign particular computational work units to particular compute nodes. The system includes a compute fabric and compute nodes that cooperatively perform a computation by mutual communication using the compute fabric. The system also includes a local data fabric that is coupled to the compute nodes, a memory, and a data node. The data node is configured to retrieve data for the computation from an external bulk data storage, and to store its work units in the memory for access by the compute nodes.

Workload management system and process

According to examples, an apparatus may include a memory to which a first queue and a second queue are assigned, in which a storage device is to access data task requests stored in the first queue and the second queue, in which the apparatus is to transfer the first queue to a second apparatus. The apparatus may also include a central processing unit (CPU), the CPU to input data task requests for the storage device into the second queue, in which the second apparatus is to store the first queue in a second memory of the second apparatus, and the storage device is to access data task requests from the first queue stored in the second memory of the second apparatus and data task requests from the second queue stored in the memory to cause the apparatus and the second apparatus to share access to the storage device.

Storage device sharing through queue transfer

The present system enables more efficient I/O processing by providing a mechanism for maintaining data within the locality of reference. One or more accelerator modules may be implemented within a solid state storage device (SSD). The accelerator modules form a caching storage tier that can receive, store and reproduce data. The one or more accelerator modules may place data into the SSD or hard disk drives based on parameters associated with the data.

Maintain data in differently performing storage devices

System method for I/O acceleration in hybrid storage wherein copies of data segments are deleted if identified segments does not meet quality level threshold

Various information about storage resources in a UNIX or UNIX derivative operating system computing environment is gathered from various sources in response to scan requests. Where a given type of information for a given storage resource is gathered from multiple sources, the information is verified for consistency, and placed in a single file in an industry standard hierarchical format. Scan threads are timed to provide reliable performance.

Storage resource scan

A configuration of a first storage resource is written to a first instance of a single file in a standard hierarchical format that is stored locally in nonvolatile memory and updatable by the first resource. A configuration of a second storage resource is written to a second instance of the single file in the standard hierarchical format stored locally in nonvolatile memory and updatable by the second resource. The first instance and second instance of the single file are updated so that all configurations are present and identical in all instances of the single file.

Redundant appliance configuration repository in standard hierarchical format

SCSI commands are used over IP (iSCSI) to trigger digest calculations of block span ranges. The parameters of the digest-calculation command include the command itself a block range, and the digest-calculation method to be used. A related combined command compares the digest-calculation result with a preexisting digest-calculation result. The parameters of the combined command include the parameters of the digest-calculation command, and the preexisting digest calculation result. Another command initializes a block range.

Generating digest for block range via iSCSI

A network storage appliance having a main computing unit is provided with an embedded microcomputer for monitoring startup and operation of the main computing unit. The microcomputer has the ability to restart, power down or power up the main computing unit if any corresponding predetermined conditions exist. Local nonvolatile storage holds the firmware for the main computing unit and configuration for the appliance. After loading the firmware and configuration to the main computing unit, the nonvolatile storage is electronically disconnected from the main computing unit. Where two such appliances are coupled for redundancy between the same network storage and the network, the embedded microcomputers are separately coupled for communication to allow for negotiating a restart of one of the main computing units under the direction of the other, or if any predetermined conditions occur.

Lightweight management and high availability controller

The portion of a source block storage resource to be replicated, and the corresponding portion of the block storage resource being written to, are each divided into a predefined number of equal-sized spans. A digest calculation is then generated for each span in a corresponding pair and compared. If the digests do not match, those spans are divided into still smaller spans, and digests are calculated and compare iteratively, to identify smaller areas of discrepancies and reduce the number of blocks that are actually written.

Growing community of inventors

Morris Plains, NJ, United States of America

Kirill Malkin