Hewlett-packard Development Company, L.p.

International Business Machines Corporation

Guillermo Alvarez

Arif Abdulhusein Merchant

Mustafa Uysal

John Wilkes

David Darden Chambliss

Prashant Pandey

Ralph Attila Becker-Szendy

Sandeep Madhav Uttamchandani

Divyesh Jadav

Jian J Xu

John Davis Palmer

Li Yin

Tzongyu Paul Lee

Ramachandran Gopalakrishna Menon

Christopher Lumb

Harry M Yudenfriend

Vijayshankar Raman

Garret Frederick Swart

Ram Swaminathan

James Alan Ruddy

Eric A Anderson

Richard Andrew Golding

Alistair Veitch

Linda Marie Duyanovich

Fabian E Bustamante

Chenyang Lu

Ted Romer

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.

Storage system including a fast storage device for storing redundant data

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.

Quality of service controller and method for a data storage system

A method is disclosed that places data-intensive subprocesses in close physical and logical proximity to the facility responsible for storing the data, so that high efficiencies at reduced cost are achieved. In one specific example, new computer programs, termed adjuncts, are added and placed in a logical partition on a storage facility so that they can be invoked using appropriate commands issued on the I/O channel. Further, programs or changes are added to existing programs on the host machine, wherein such programs or changes discover the function extensions and invoke them to perform data processing.

System to delegate virtual storage access method related file operations to a storage server using an in-band RPC mechanism

A system for utilizing informed throttling to guarantee quality of service to a plurality of clients includes a server core having a performance analyzer that compares a performance level received by a client to a corresponding contracted service level and determines if the client qualifies as a victim whose received performance level is less than the corresponding contracted service level. The performance analyzer is further configured to identify one or more candidates for throttling in response to an I/O stream receiving insufficient resources by determining if the client qualifies as a candidate whose received performance level is better than the corresponding contracted service level. The server core further includes a scheduler that selectively and dynamically issues a throttling command to the candidate client, and provides a quality of service enforcement point by concurrently monitoring a plurality of I/O streams to candidate clients and concurrently throttling commands to the candidate clients.

Utilizing informed throttling to guarantee quality of service to I/O streams

A technique for emulation of a data storage system. The invention allows the level of services to be provided by a data storage system to be specified in terms of the level of services provided by another storage system. In one aspect, a performance characterization of a data storage device to be emulated is obtained (e.g., by experimental techniques). A specification of a workload is also obtained that includes a specification of a plurality of data stores for the workload. The data stores are assigned to an emulation data storage device according to the performance characterization and according to the specification of the workload such that sufficient resources of the emulation data storage device are allocated to the workload to meet the performance characterization of the data storage device to be emulated. The emulation data storage device is then operated under the workload. Quality-of-service (QoS) control may be performed so as to provide a degree of performance isolation among the workloads.

Data storage system emulation

A technique of allocating shared resources in a computer network-based storage system comprises taking periodic performance samples on a running storage system; evaluating an objective function that takes as input the performance samples to quantify how aligned a current state of the storage system is with organizational objectives; building and maintaining models of behavior and capabilities of the storage system by using the performance samples as input; determining how resources of the storage system should be allocated among client computers in the storage system by selecting one among many possible allocations based on predictions generated by the models in order to maximize a value of the objective function; calculating a confidence statistic value for a chosen resource allocation based on an accuracy of the models; and enforcing the chosen resource allocation on the running storage system when the confidence statistic value is at or above a predetermined threshold value.

Approach based on self-evolving models for performance guarantees in a shared storage system

The present system and associated method resolve the problem of providing statistical performance guarantees for applications generating streams of read/write accesses (I/Os) on a shared, potentially distributed storage system of finite resources, by initiating throttling whenever an I/O stream is receiving insufficient resources. The severity of throttling is determined in a dynamic, adaptive way at the storage subsystem level. Global, real-time knowledge about I/O streams is used to apply controls to guarantee quality of service to all I/O streams, providing dynamic control rather than reservation of bandwidth or other resources when an I/O stream is created that will always be applied to that I/O stream. The present system throttles at control points to distribute resources that are not co-located with the control point. A competition model is used with service time estimators in addition to estimated workload characteristics to determine which I/O needs to be throttled and the level of throttling required. A decision point issues throttling commands to enforcement points and selects which streams, and to what extent, need to be throttled.

System and method for utilizing informed throttling to guarantee quality of service to I/O streams

A technique for programmatically obtaining experimental measurements for model construction. A user provides criteria for the model, such as computational algorithms which characterize behavior of the real system, specifications of experiments to be performed on the real system for collecting experimental data from the real system, an identification of sought parameters which are to be derived from results of the experiments and desired tolerance constraints on the sought parameters. From experimental data collected from the real system and from the provided criteria, the inventive method and apparatus programmatically determines in an iterative loop which additional experiments are to be performed in order to achieve the desired tolerance constraints. After one or more iterations of the loop, the values for the sought parameters are determined within the desired tolerance constraints.

Technique for programmatically obtaining experimental measurements for model construction

Embodiments herein present a method, system, computer program product, etc. for automated management using a hybrid of prediction models and feedback-based systems. The method begins by calculating confidence values of models. Next, the method selects a first model based on the confidence values and processes the first model through a constraint solver to produce first workload throttling values. Following this, workloads are repeatedly processed through a feedback-based execution engine, wherein the feedback-based execution engine is controlled by the first workload throttling values. The first workload throttling values are applied incrementally to the feedback-based execution engine, during repetitions of the processing of the workloads, with a step-size that is proportional to the confidence values. The processing of the workloads is repeated until an objective function is maximized, wherein the objective function specifies performance goals of the workloads.

Self-modulation in a model-based automated management framework

A computer storage system includes a controller and a storage device array. The storage device array may include a first sub-array and a fast storage device sub-array. The first sub-array includes one or more first storage devices storing data. The fast storage device sub-array includes one or more fast storage devices storing a copy of the data stored in the first sub-array.

Storage system using fast storage device for storing redundant data

Data is migrated by moving a set of the data in a data storage system of a computer system. The method includes monitoring a performance of at least one executing application while moving the data in the data storage system. A change in the rate of moving is calculated in response to the monitored performance of the executing application. The rate of moving is modified in accordance with the calculated change.

Method and system for online data migration on storage systems with performance guarantees

A computer storage system includes a controller and a storage device array. The storage device array includes a first sub-array and a fast storage device sub-array. The first sub-array includes one or more log-structured storage devices storing data. The fast storage device sub-array includes one or more fast storage devices storing a copy of the data stored in the first sub-array.

Storage system using fast storage and log-structured storage

A method predicts performance of a system that includes a plurality of interconnected components defining at least one data flow path. The method references a workload specification for the system. The method models the system using one or more component models. Each component model represents selected one or more of the components. Each component model is arranged in like relationship to the data flow path as the selected one or more of the components represented by the component model. Each component model is (a) a constraint upon the workload specification input to that component model or (b) a transformer of the workload specification input to that component model so as to result in one or more output workload specifications that are input workload specifications to subsequent component models along the data flow path or (c) both a constraint and a transformer. At least one of the component models is a constraint. At least some of the component models along the data flow path operate on the workload specification. In one preferred form, operating on the workload specification involves arranging the component models in a hierarchy corresponding to the data flow path; using the specified workload specification as input to the topmost component model in the hierarchy; and applying one or more of the component models to its input workload specification, starting with the topmost component model and then component models at progressively lower levels in the hierarchy. Output workload specification at one level is input workload specification at the next lower level. If the component model comprises a constraint, the method evaluates whether the input workload specification satisfies or violates the constraint. If the component model comprises a workload specification transform, the method modifies the input workload specification so as to produce one or more output workload specifications.

Method and apparatus for morphological modeling of complex systems to predict performance

A method of and apparatus for determining whether a multi-component target system meets a given multi-part performability requirement is provided. A description of the target system, failure probabilities for components of the target system and a multi-part performability requirement for the target system are obtained. The multi-part performability requirement indicates desired performance levels and corresponding fractions of time. One or more failure-scenarios are successively computed that represent one or more states of the target system having zero or more components failed and a corresponding probability of occurrence of the one or more of the states of the target system. Performance of the target system is modeled under the failure scenarios using a performance predictor module for generating a multi-part performability function. The multi-part performability function is compared with said multi-part performability requirement to determine whether the target system meets the multi-part performability requirement.

Growing community of inventors

San Jose, CA, United States of America

Guillermo Alvarez