International Business Machines Corporation

Bruce McNutt

Joseph S Hyde, Ii

Vernon Walter Miller

Yan Xu

Yong Guo

Thomas Charles Jarvis

Michael T Benhase

Dharmendra S Modha

Binny Sher Gill

Jie Tian

John G Aschoff

Richard Anthony Ripberger

David Darden Chambliss

Eduardo Campos Peiro

Joshua Douglas Martin

Neena A Cherian

Tao Tang

Brian J Smith

Ruth Enid Azevedo

Matthew J Kalos

Jaishankar Moothedath Menon

David Alan Burton

Barry Douglas Whyte

Gary Eugene Morain

Jeffrey Allen Berger

Steven Robert Lowe

Kevin F Smith

Barrie N Harding

Stanley C Kurtz

Stuart Robert Goodgold

David Alastair Short

A computer implemented method including receiving a set of utilization metrics for a system comprising at least an average number of concurrent requests to the system and a maximum concurrency that the system is capable of supporting, providing a function that incorporates two curve segments, computing a utilization according to a ratio of the average concurrent requests to the function, and managing performance problems indicated by the utilization. A computer implemented method including receiving a set of response time metrics comprising at least an average response time, average concurrent requests, and a minimum interference response time, computing a current response ratio of the minimum interference response time and the average response time, computing a maximum response ratio corresponding to a maximum concurrency, determining the maximum concurrency is inaccurate by comparing the maximum response ratio and the current response ratio, and replacing the maximum concurrency.

System resource component utilization

A computer-implemented method according to one embodiment includes identifying aggregated customer performance data for a system resource, creating a first system map for the system resource, utilizing the aggregated customer performance data, comparing the first system map to a second system map created for the system resource utilizing calibration data, and adjusting the second system map, based on the comparing.

Storage system performance models based on empirical component utilization

According to one embodiment, a computer-implemented method includes obtaining, during a measurement period, a set of utilization metrics for a system; determining, whether the average number of outstanding system requests of the system lies between a current and an actual value of the maximum concurrency; in response to determining that the average number of outstanding system requests of the system does not lie between the current and actual values of the maximum concurrency, not adjusting the maximum concurrency value; in response to determining that the average number of outstanding system requests of the system does lie between the current and actual values of the maximum concurrency, assigning a value of the maximum concurrency to be the value of the average number of outstanding system requests; accessing a utilization function; solving the utilization function; and managing performance problems of the system indicated by solving the utilization function.

Maintaining manageable utilization in a system to prevent excessive queuing of system requests

A computer-implemented method according to one embodiment includes receiving, by one or more processors, a set of utilization metrics for a system comprising at least an average number of concurrent requests N to the system and a maximum concurrency c that the system is capable of supporting, computing, by one or more processors, a tipping point p, computing, by one or more processors, a utilization U according to a ratio of the average number of concurrent requests N to a function, where the function is defined as: c+f(f(N/p)) for N&#x3c;p, and N+1+b*f(p/N) for N&#x2265;p, where b=c&#x2212;p and f(y)=y&#x2212;b*y*(1&#x2212;y)+0.5*b*(b&#x2212;1)*y*(1&#x2212;y), and adjusting, by one or more processors, the system, utilizing the utilization U.

Various embodiments for managing data by a processor in a multi-tiered computing storage environment. Based on Input/Output (I/O) statistics, a ranking mechanism is applied to differentiate data between at least a cache rank and a Solid State Drive (SSD) rank. The ranking mechanism migrates data between the cache rank and SSD rank such that storage space in the cache rank is reserved for those of the plurality of data workload types having a greater adverse effect on a storage performance characteristic if stored in the SSD rank than if those workload types were stored in the cache rank.

Storage in tiered environment with cache collaboration

Monitoring the division of input/output (I/O) loading within a storage system is provided where the storage system includes one or more storage devices and one or more ranks, each rank including a plurality of extents. The I/O loading of individual extents is estimated for each of the one or more ranks of the storage system by determining an average disk utilization in a rank, measuring a total heat for all extents in the rank, and based on the determined average disk utilization in the rank and the measured total heat for all extents in the rank, calculating a contribution of each extent to the utilization of the rank.

Estimation of performance utilization of a storage device

Monitoring a level of utilization is provided. A threshold is determined based, at least in part, on a count of service channels of a device. An upper boundary value of a numerical range is determined based, at least in part, on the count of service channels. A lower boundary value of the numerical range is determined based, at least in part, on the threshold. Whether a count of outstanding requests of the device is contained within the numerical range is determined. An estimated level of utilization of the device is determined based, at least in part, on the upper boundary value, the lower boundary value, the count of service channels, and the count of outstanding requests. The estimated level of utilization is reported.

Monitoring levels of utilization of device

Monitoring a level of utilization is provided. An initial numerical range based, at least in part, on a count of service channels of a device is determined. A candidate numerical range, defined by an upper value and a lower value, based, at least in part, on the initial numerical range, is determined. A level of utilization of a first measurement interval of the device is estimated by: repeatedly updating the lower value and the upper value based, at least in part, on the level of utilization, until the lower value and the upper value differ less than a pre-determined threshold; and determining an estimated level of utilization based, at least in part, on the lower value and the upper value. The estimated level of utilization is reported.

Monitoring device usage

Various embodiments for managing data by a processor in a multi-tiered computing storage environment. Input/Output (I/O) statistics are examined from each of cache and device drivers in the computing storage environment. Based on the I/O statistics, a ranking mechanism is applied to differentiate data between at least a cache rank and a Solid State Drive (SSD) rank. The ranking mechanism migrates data between the cache rank and SSD rank such that storage space in the cache rank is reserved for those of the plurality of data workload types having a greater adverse effect on a storage performance characteristic if stored in the SSD rank than if those workload types were stored in the cache rank.

A method, for migrating newly allocated data into a second storage tier, includes receiving a request to store a data item. The method includes allocating an extent to a first storage tier to store the new data item. The method includes determining whether an extent is newly allocated on the first storage tier. The method includes migrating, by avoiding a learning phase, the newly allocated extent to the second storage tier from the first storage tier.

Migration of newly allocated data to a storage tier

Method, system, and computer program product embodiments for facilitating deduplication product testing in a computing environment are provided. In one such embodiment, data to be processed through the deduplication product testing is arranged into a single, continuous stream. At least one of a plurality of random modifications are applied to the arranged data in a self-similar pattern exhibiting scale invariance. A plurality of randomly sized subsets of the arranged data modified with the self-similar pattern is mapped into each of a plurality of randomly sized deduplication test files which are calibrated against input/output (I/O) trace data obtained in the computing environment.

Generation of realistic file content changes for deduplication testing

Estimating a level of concurrency is provided. An estimated level of concurrency of a device is determined. A first average response time, wherein the first average response time is an average of response times of a first set of operations of the device is determined. A second average response time is determined, wherein the second average response time is an average of response times of a second set of operations of the device, wherein each of the second set of operations is initiated under conditions of minimal interference of the device. A threshold based on the estimated level of concurrency is determined. The estimated level of concurrency is updated based, at least in part, on a ratio of the second average response time to the first average response time.

Free space collection in a log structured storage system. In response to receiving a request to collect free space, the segment in a recently invalidated segment list having the largest cumulative sum of invalidation duration values is selected for free space collection if this cumulative sum is larger than the cumulative sum of invalidation duration values for the segment in a stable segment list having the largest cumulative sum of invalidation duration values. Otherwise, the segment in the stable segment list having the largest cumulative sum of invalidation duration values is selected for free space collection. emptying. The selected segments are then emptied and become available for the writing of data to their tracks.

Growing community of inventors

Tucson, AZ, United States of America

Bruce McNutt