Abstract: An embodiment of a method of cooperative caching for a distributed storage system begins with a step of requesting data from storage devices which hold the data. The method continues with a step of receiving any cached blocks and expected response times for providing non-cached blocks from the storage devices. The method concludes with a step of requesting a sufficient number of the non-cached blocks from one or more particular storage devices which provides an expectation of optimal performance.

Abstract: A system and method for allocating resources on a shared storage system are disclosed. The system 10 can include a shared storage device 12 and a plurality of port schedulers 14 associated with a plurality of I/O ports 16 that are in communication with the shared storage device 12. Each port scheduler 14 is configured to enforce a concurrency level and a proportional share of storage resources of the shared storage device 12 for each application 18 utilizing the associated port. The system 10 can also include a resource controller 17 that is configured to both monitor performance characteristics of the applications 18 utilizing at least one of the I/O ports 16, and to adjust the concurrency level and the proportional share of storage resources parameters of the port schedulers 14 for at least a portion of the applications 18 in order to vary allocation of the resources of the shared storage device 12.

Abstract: A system comprises a processor and an alert module coupled to the processor. The processor the processor monitors performance data; determines a subset of the performance data, the subset correlated with a measure of underperformance; determines a trend of the subset, the trend correlated with the measure; and identifies an occurrence of the trend. The alert module outputs an alert based on the identification.

Abstract: To determine files associated with one or more workflows, a trace of accesses of files in at least one server is received. The files are grouped into at least one set of files, where the files in the set are accessed together more than a predetermined number of times in the trace. Files associated with the particular workflow are identified based on the at least one set.

Abstract: A method and a system of improving recoverability of a dataset associated with a multi-tier storage system are provided. Updates to a dataset are stored in first storage. The dataset and a point-in-time copy of the dataset reside on second storage, and the first storage provides higher reliability than the second storage. The dataset is restored using at least a subset of the stored updates in combination with the point-in-time copy in the event that the dataset is lost.

Abstract: A computer storage system includes a controller, a first storage device and a second storage device including at least one fast storage device. The controller is configured to perform data operations. The first storage device stores data, and the second storage device stores data redundant to the data stored in the first storage device.

Abstract: A system includes disk storage to provide differentiated storage QoS for a plurality of IO classes. Each IO class has a plurality of applications to it. A QoS controller collects IO statistics for each application and each class. The QoS controller adaptively determines an IO class assignment for each application based at least in part on the collected IO statistics.

Abstract: A quality-of-service controller and related method for a data storage system. Requests for each of a plurality of storage system workloads are prioritized. The requests are selectively forwarded to a storage device queue according to their priorities so as to maintain the device queue at a target queue depth. The target queue depth is adjusted response to a latency value for the requests wherein the latency value is computed based on a difference between an arrival time and a completion time of the requests for each workload. Prioritizing the requests may be accomplished by computing a target deadline for a request based on a monitored arrival time of the request and a target latency for its workload. To reduce latencies, the target queue depth may be reduced when the target latency for a workload is less than its computed latency value. To increase throughput, the target queue depth may be increased when the target latency for each workload is greater than each computed latency value.

Abstract: A model is constructed to determine performance of each computing application based on allocation of resources (including at least one hardware resource) to the computing applications. How the allocation of the resources to the computing applications affects the performance is unknown beforehand. The resources are allocated to the computing applications based at least on the model. Where the resources are overloaded as allocated to the computing applications, performance degradation of each computing application is performed based at least on priorities of the computing applications relative to one another and on the model. Performance degradation reduces usage of the resources by the computing applications so that the resources are no longer overloaded. How the priorities of the computing applications affect the performance degradation in a relative manner to one another is known and predictable beforehand.

Abstract: An embodiment of a method of writing data includes issuing write messages to a replica set of storage devices. Write confirmations are received from at least a majority of the storage devices. An embodiment of a method reading data includes issuing read messages to a replica set of storage devices. Read confirmations are received from at least a first majority of the storage devices. Read commit messages are issued to the storage devices. Commit confirmations are received from at least a second majority of the storage devices.

Abstract: An embodiment of a method of hashing an address space to a plurality of storage servers begins with a first step of dividing the address space by a number of the storage servers to form data segments. Each data segment comprises a base address. A second step assigns the data segments to the storage servers according to a sequence. The method continues with a third step of measuring a load on each of the storage servers. According to an embodiment, the method concludes with a fourth step of adjusting data shares assigned to the storage servers according to the sequence to approximately balances the loads on the storage servers while maintaining the base address for each data segment on an originally assigned storage server. According to another embodiment, the method periodically performs the third and fourth steps to maintain an approximately balanced load on the storage servers.

Abstract: One embodiment of the present invention provides a method comprising assigning a first weight to a first consumer and a second weight to a second consumer. The method further comprises receiving at a first scheduler a request for service by a first resource to the first consumer; receiving at a second scheduler a request for service by a second resource to the first consumer; and receiving at the first scheduler a request for service by the first resource to the second consumer. The method further comprises scheduling, by the first and second schedulers, service from the first and second resources to the first and second consumers, wherein the scheduling maintains total service provided by the first and second resources to the first and second consumers in proportion to the respective weights of the first and second consumers without requiring the first and second schedulers to communicate with each other.

Abstract: Requests that are executing when an application is determined to be in an overload condition are preempted. To select the executing requests to preempt, a value for each executing request is determined. Then, executing requests are selected for preemption based on the values.

Abstract: Read, write and recovery operations for replicated data are provided. In one aspect, a system for redundant storage of data included a plurality of storage devices and a communication medium for interconnecting the storage devices. At least two of the storage devices are designated devices for storing a block of data. Each designated device has a version of the data and a first timestamp that is indicative of when the version of data was last updated and a second timestamp that is indicative of any pending update to the block of data. The read, write and recovery operations are performed to the data using the first and second timestamps to coordinate the operations among the designated devices.

Abstract: 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.

Abstract: 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.

Abstract: Embodiments include methods, apparatus, and systems for prefetching data in distributed storage systems. One method of software execution includes using input/output (I/O) requests from multiple separate networked storage nodes in a distributed storage system to prefetch data for one of the multiple separate networked storage nodes.

Abstract: To determine files associated with one or more workflows, a trace of accesses of files in at least one server is received. The files are grouped into at least one set of files, where the files in the set are accessed together more than a predetermined number of times in the trace. Files associated with the particular workflow are identified based on the at least one set.

Abstract: To schedule workloads of requesters of a shared storage resource, a scheduler specifies relative fairness for the requesters of the shared storage resource. In response to the workloads of the requesters, the scheduler modifies performance of the scheduler to deviate from the specified relative fairness to improve input/output (I/O) efficiency in processing the workloads at the shared storage resource.

Abstract: A configuration method includes selecting a design configuration, and modeling the solution configuration having a failure event for the selected solution configuration to determine an amount of down time associated with the failure event. A first penalty cost function is applied the amount of down time to determine a first penalty cost associated with the failure event of the selected solution configuration.