Hewlett-packard Development Company, L.p.

Compaq Computer Corporation, Inc.

James Michael Reuter

David William Thiel

Richard Fitzhugh Wrenn

Robert George Bean

Alistair Veitch

James S Jackson

Arif Abdulhusein Merchant

Svend Frolund

Leslie B Lamport

Andrew C St Martin

Eliezer Gafni

Douglas L Voigt

Richard Franklin Lary

Richard Perham Helliwell

Susan Spence

Rodger Daniels

Christopher Stroberger

Charles E Christian, Jr

Michael Padovano

Michael G Byrnes

Lukas Lloyd Wardensky

Laura Clemens

Donald R Fraser

Deborah Levinson

In a storage system, at least three storage sites are arranged in a full mirroring relationship using a virtual connection between instances of a storage group on two of the sites whereby no user data is transmitted.

Data mirroring using a virtual connection

Various embodiments of the present invention are directed to a distributed lock and distributed locking protocol to allow multiple communicating entities to share access to a computing resource. Certain embodiments of the present invention employ a data storage register implemented by the communicating entities to hold a value reflective of a distributed lock state.

Distributed lock

Various embodiments of the present invention are directed to distributed data-storage systems that provide non-distributed-data-storage-device-like interfaces to host computers that access the distributed data-storage systems. Distributed data-storage systems generally rely on state information used for execution of commands, received from host computers, by component data-storage systems. Command-executing component data-storage systems may access and modify state information shared among many or all of the component data-storage systems. Embodiment methods of the present invention provide methods for partitioning state information within a distributed data-storage system as well as communications protocols and methods for managing shared state information among the component data-storage systems within a distributed data-storage system. In certain embodiments of the present invention, state information is partitioned into one or more of: (1) local state information that is managed, accessed, and modified separately by each component data storage system; (2) shared state information that is locally cached on component data storage systems for immediate read access, that is periodically refreshed, but that is kept globally consistent among component data storage systems by distributing state-modifying operations; and (3) continuously consistent, shared state information.

Distributed-state-information-based distributed computing systems and methods and protocols for managing distributed state information

A system includes a volatile storage, a persistent storage, a capacitor-based power supply, and a controller coupled to the capacitor-based power supply. The controller detects interruption of main power, and in response to detecting the interruption of main power, begins backup copying of data from the volatile storage to the persistent storage. After beginning the backup copying of data, the controller checks whether the main power has resumed prior to depletion of the capacitor-based power supply. In response to detecting that main power has resumed prior to depletion of the capacitor-based power supply, the controller resumes operation using content of the volatile storage without restoring data from the persistent storage.

Providing a capacitor-based power supply to enable backup copying of data from volatile storage to persistent storage

A unit of data is read from memory of a replicated cache. The replicated cache comprises a plurality of independent computing devices. Each independent computing device comprises a processor and a portion of the memory. Confirmations are received from at least a majority of the independent computing devices that a flush operation for the unit of data was initiated no later than a time indicated by the timestamp and that a more recent version of the unit of data has not been flushed. The unit of data is provided to storage.

Operating a replicated cache that includes receiving confirmation that a flush operation was initiated

Data blocks are read from a distributed cache. The distributed cache comprises m replicated caches, each replicated cache including a plurality of independent computing devices. Each independent computing device of the replicated caches holds a replica of a particular one of the m data blocks in memory. The m data blocks and p parity blocks are stored across m plus p independent computing devices. Each of the m plus p independent computing devices stores a single block selected from the m data blocks and the p parity blocks.

Method of operating distributed storage system in which data is read from replicated caches and stored as erasure-coded data

A virtual storage system comprises an agent to store a first table to map virtual storage segments to storage locations, and a controller to store a second table and to intermittently cause contents of the first table to be replaced by contents of the second table. During an I/O operation, one of the entries in the first table can be accessed by a host to determine one of the storage locations. Also, a method for performing an I/O operation on a virtual storage segment, defined by first table that maps a storage segment to a first storage location, comprises turning off I/O operations at the first storage location, identifying portions of the virtual storage segment to be effected during a write operation, storing a record of identified portions at a second table not the first table, and writing to portions of the second storage location associated with the identified portions.

Centralized fine-grained enhancements for distributed table driven I/O mapping

Embodiments of the present invention are directed to digitally encoded hierarchical timestamps that are stored in computer readable mediums. The hierarchical timestamps of the present invention include a first field that stores a timestamp value and a second field that stores a processing-level indication that indicates the scope of the timestamp within a hierarchical processing system.

Hierarchical timestamps

The present invention provides a method for copying data through a virtualized storage system using distributed table driven (I/O) mapping. In a system having a virtual disk (the 'original disk'), a persistent mapping table for this virtual disk exists on a controller, and volatile copies of some or all entries in this mapping table are distributed to one or more more mapping agents. The method of the present invention creates a new virtual disk mapping table that has the exact same entries as the mapping table as the original virtual disk. The new snapshot disk then shares the same storage as the original disk, so it is space efficient. Furthermore, creating new snapshot disk involves only copying the contents of the mapping table, not moving data, so the creation is fast. In order to allow multiple virtual disks to share storage segments, writes to either the original virtual disk or the snapshot copy cannot be seen by the other. Therefore, in addition to simply copying the mapping table, both the original and snapshot disk mapping table must also cause writes to these disks to be handled specially. Finally, any changes to the original disk mapping table stored in the controller must be coordinated with the volatile copies stored in the mapping agent in such a way so that all hosts see a consistent view of that disk.

Generating updated virtual disks using distributed mapping tables accessible by mapping agents and managed by a centralized controller

The present invention provides a system and method for creating virtualized storage in a storage area network using distributed table-driven input/output mapping. The present invention distributes the virtualization mapping in multiple parallel, mapping agents that are separate from a controller. This allows the performance-sensitive mapping process to be parallelized and distributed optimally for performance, while the control of the mapping may be located in a controller chosen for optimal cost, management, and other implementation practicalities. The mapping agents store the virtual mapping tables in volatile memory, substantially reducing the cost and complexity of implementing the mapping agents. The controller is responsible for persistent storage of mapping tables, thereby consolidating the costs and management for persistent mapping table storage in a single component. Distributed virtualization also allows the controller to manage multiple virtual disks used by multiple host systems, and allows a single virtual disk to be shared by multiple host systems. The mapping agents preferably do not interact only with other mapping agents, thereby improving the scalability of the virtual storage system and the virtual storage system's tolerance of component failures.

Architecture for parallel distributed table driven I/O mapping

Briefly, the present invention provides a system and method for distributing SCSI semantics throughout a network. Specifically, the present invention distributes the SCSI semantics through multiple parallel agents and a separate controller. This configuration allows performance-sensitive distribution of SCSI semantics to be parallelized and optimized for performance in the agents, while the control and management of the SCSI semantics is centralized in the controller chosen for optimal cost, management, and other implementation practicalities. In this design, the SCSI semantics are stored in the controller, and portions of the SCSI semantics are distributed to the mapping agents as cached, read-only information. The controller is responsible for persistent storage of the SCSI semantics, thereby consolidating the costs and management for the SCSI semantics in a single component. The agents preferably do not interact with other mapping agents, thereby improving the scalability of virtual storage systems and their tolerance of component failures. Within an embodiment of the SCSI distribution system, key SCSI semantic information is distributed to the agents while infrequently used semantic information is centralized in the controller. Likewise, SCSI commands that cannot be processed by the agents may be forwarded to the controller.

Structure and process for distributing SCSI LUN semantics across parallel distributed components

The present invention provides a virtual storage system that generally stores uses larger segmentations, but divides large segments into smaller sub-segments during data movement operations. The present invention provides a method and system having this hierarchy of segment sizes, namely a large segment for the normal case, while breaking the large segment into single disk blocks during data movement. The mapping has large segments except for those segments undergoing data movement. For those segments, it would be desirable to have the smallest segment size possible, namely, a single disk block. In this way, the administration costs are generally low, but latencies caused by the movement of large data blocks are avoided.

Distributed fine-grained enhancements for distributed table driven I/O mapping

A system for moving physically stored data in a distributed, virtualized storage network is disclosed. A group of data sets is written to a first storage device as part of a write operation such as migration. A plurality of storage devices partially filled with data are designated as substitutes. The write operation to the first storage device is suspended upon receiving a request to read a data set stored in the first storage device, such as occurs in a recall operation. A second storage device is then selected from the plurality of substitute storage devices. The write operation is continued by writing data sets from the group of data sets included in the write operation that were not written to the first storage device to the selected second storage device. The requested data is then read from the first storage device. After data has been read from the first storage device, the first storage device may be designated as a substitute storage device so that the partially filled first storage device may be selected for continuing write operations. Data sets from substitute storage devices may be transferred or merged into a lesser number of storage devices during recycle operations to prevent the number of substitute storage devices from increasing beyond a predetermined limit or goal. Recycling operations in which data sets from different storage devices are transferred or merged may be performed by building a first queue including a list of filled tapes ordered according to the least amount of valid data and a second queue including all unassociated partially filled storage devices ordered by the amount of available storage space, and merging.

Growing community of inventors

Colorado Springs, CO, United States of America

James Michael Reuter