Patent application number | Description | Published |
20100312959 | SPATIAL LOCALITY OF FILE SYSTEM BLOCK ALLOCATIONS FOR RELATED ITEMS - Example apparatus, methods, and computers facilitate improving spatial locality of file system block allocations for related items. One example method includes accessing a block allocation request that is configured to control a computer to allocate a block of storage available on a storage device and, upon determining that the block allocation request includes a session key that identifies a session associated with a block reservation pool, controlling the computer to allocate a block of storage from the block reservation pool to satisfy the block allocation request rather than allocating a block of storage from generally available blocks of storage. The block reservation pool includes blocks of storage on the storage device that satisfy a spatial locality relationship determined by a pre-read technique associated with the storage device. | 12-09-2010 |
20110107139 | TIMER BOUNDED ARBITRATION PROTOCOL FOR RESOURCE CONTROL - Example apparatus, methods, and computers prevent a split brain scenario in a pair of high availability servers by maintaining single writer access to a resource by controlling the resource according to a timer bounded arbitration protocol that controls self-termination of a writer process. One example method includes monitoring control of an arbitration (ARB) block by an active file system manager (FSM) and selectively causing a selection of a standby metadata controller (MDC) when control of the ARB block does not satisfy the timer bounded ARB protocol. The example method also includes selectively forcing a hardware reset of an apparatus running the active FSM and selectively establishing an FSM on a selected redundant MDC as a replacement FSM. | 05-05-2011 |
20110113066 | ADAPTING A TIMER BOUNDED ARBITRATION PROTOCOL - Example apparatus, methods, and computers prevent a split brain scenario in a pair of high availability servers by maintaining single writer access to a resource. One example method includes monitoring control of an arbitration (ARB) block by an active file system manager (FSM). An active file system portmapper (FSMPM) simultaneously requests permission from a standby FSM to restart a dead-man timer. The active FSM attempts to maintain control of the ARB block before expiration of the dead-man timer. If the active FSM is unable to maintain control of the ARB block before expiration of the dead-man timer, a hardware reset of an apparatus running the active FSM is forced. Therefore, the active FSMPM and standby FSM negotiate for additional time to maintain control of the ARB block to maintain single writer access and avoid unnecessary hardware resets. | 05-12-2011 |
20120310999 | Adapting A Timer Bounded Arbitration Protocol - Example apparatus, methods, and computers prevent a split brain scenario in a pair of high availability servers by maintaining single writer access to a resource. One example method includes monitoring control of an arbitration (ARB) block by an active file system manager (FSM). An active file system portmapper (FSMPM) simultaneously requests permission from a standby FSM to restart a dead-man timer. The active FSM attempts to maintain control of the ARB block before expiration of the dead-man timer. If the active FSM is unable to maintain control of the ARB block before expiration of the dead-man timer, a hardware reset of an apparatus running the active FSM is forced. Therefore, the active FSMPM and standby FSM negotiate for additional time to maintain control of the ARB block to maintain single writer access and avoid unnecessary hardware resets. | 12-06-2012 |
20150227416 | Combined Asynchronous and Synchronous Fountain Code Storage In an Object Store - Example apparatus and methods produce a set of rateless erasure codes (e.g., fountain codes) for a file stored in a primary data store (e.g., hard drive) or in an archive system. The archive system may store the file in a redundant array of independent disks (RAID). A first subset of the rateless erasure codes are stored in an object storage using a synchronous protocol. A second subset of rateless erasure codes are stored in the object storage using an asynchronous protocol. The object storage system may inform the archive system when desired redundancy has been achieved or when desired redundancy has been lost. The archive system may buffer rateless erasure codes before providing the codes to the object storage to improve performance. A failure in the archive system or object storage system may be mitigated by retaining the file in the primary data store until the desired redundancy is achieved. | 08-13-2015 |
20150227423 | Mitigating The Impact Of A Single Point Of Failure In An Object Store - Example apparatus and methods distribute ranges or erasure codes associated with ranges to reduce or minimize the impact of a single point of failure in an object store. Erasure codes associated with related ranges to be stored in an object store may be accessed and selectively distributed to different storage devices associated with the object store. The erasure codes may be distributed according to a distribution plan so that an unavailability of one storage device will cause less than all of the related ranges to become unavailable. Example apparatus and methods may also provide a partial GET operation that will retrieve erasure codes associated with less than an entire object or with less than all possible ranges for an object. The partial GET operation may facilitate reconstructing less than an entire object, which may be valuable in, for example, weblog analytics. | 08-13-2015 |
20150229338 | Dynamically Reconfiguring Rebuild Agents For Erasure Code Rebuild - Example apparatus and methods support reconstructing an item from a set of erasure codes (e.g., fountain codes). Information about computer resources available to support reconstructing the item may be accessed and analyzed to control the spawning of multiple computer processes to support reconstructing the item. The information may include, for example, utilization and capacity data. The information may be provided by sensor agents that monitor the resources. The multiple computer processes may operate at least partially in parallel. Resources may be identified and computer processes may be spawned until all the resources that can contribute to the reconstruction are used. In one embodiment, computer processes may be spawned until the marginal utility of spawning another process falls below a threshold. | 08-13-2015 |
20150229339 | Adjusting Redundancy In An Erasure Code Object Store To Account For Varying Data Value - Example apparatus and methods control the number of rateless erasure codes (e.g., fountain codes) stored in an object store for an item (e.g., file stored as object). The codes for the item may be generated according to an M/N policy. A first safety factor that controls how many codes are stored initially in the object store is identified. A first number of codes are then stored in the object store, where the first number is selected as a function of the first safety factor. A second safety factor for the item and a condition under which the second safety factor is to be used to control the number of codes to be stored in the object store is also identified. When the condition is detected, a second number of codes are stored in the object store, where the second number is selected as a function of the second safety factor. | 08-13-2015 |