Emc IP Holding Company LLC

Dell Products Inc.

Emc Corporation

Owen Martin

Arieh Don

Malak Alshawabkeh

Adnan Sahin

Ramesh Doddaiah

Hui Wang

Sean C Dolan

Xiaomei Liu

Marik Marshak

Alexandr Veprinsky

John Creed

Benjamin Randolph

Michael J Scharland

Amnon Naamad

Alex Veprinsky

Vladimir Desyatov

Dustin Hunt Zentz

Dan Aharoni

Jeffrey Wilson

Jaeyoo Jung

Sanjib Mallick

Philip Derbeko

Jeremy O'Hare

Alesia Tringale

Earl Medeiros

Peter Blok

Mark D Moreau

Vinay G Rao

Parmeshwr Prasad

Rômulo Teixeira De Abreu Pinho

Stephen Smaldone

Sachin More

Jonas Furtado Dias

Shaul Dar

Michael E Specht

Gabriel Benhanokh

Ahmet Kirac

Paras Pandya

Philippe Armangau

Vamsi K Vankamamidi

Mark J Halstead

Rahul Deo Vishwakarma

Benjamin W Yoder

Erik P Smith

Tiago Salviano Calmon

Kuolin Hua

Felix Shvaiger

Vinícius Michel Gottin

Adriana Bechara Prado

Vasudevan Subramanian

Krishna Deepak Nuthakki

Andrew Chanler

Eduardo Vera Sousa

Igor Achkinazi

Anoop Raghunathan

Peter Beale

Robert DeCrescenzo

Lev Knopov

Diego Salomone Bruno

James L Davidson

Fatemeh Azmandian

Kaustubh S Sahasrabudhe

Rasa Raghavan

Abhilash Sanap

Benjamin A F Randolph

Peter Linden

Ken Dorman

Sichao Zhu

Thomas F O'Neill

Hani Zaitoun

Scott J Romano

Sean B Reilly

Benjamin Allen Fitz Randolph

William R Smith-Vaniz

Christopher G LeClair

Mario Würzl

Storage arrays include inline compression hardware that can simultaneously implement multiple compression levels at line rate. For each compression level, compression efficiency is inversely related to processing efficiency. A compression level is dynamically selected for segments of replication data based on one or more of compression hardware utilization, network utilization, network latency, data compressibility, and IO size. The compression hardware utilization may be maintained at or near full utilization. Extents of data of a replica are analyzed based on compressibility. Replication data that resides in an extent of relatively incompressible data, or that is associated with a relatively small IO, may be compressed using a compression level characterized by greater processing efficiency.

Dynamic compression for data replication

One or more aspects of the present disclosure relate to dynamic performance optimization in storage arrays using tracelet matrices. In embodiments, one or more input/output (IO) operations are received by a storage array. Code path branches taken by each IO operation across storage array components are recorded in a bitmap matrix. Additionally, the likelihood of different code path branches being taken by each IO operation at a storage device or storage group level is forecasted using each IO operation's corresponding bitmap matrix. Performance bottlenecks corresponding to the one or more IO operations are also identified based on a comparison of each IO operation's bitmap matrix and a forecast of the likelihood of the different code path branches being taken by each IO operation targeting a subject storage device or storage group. Further, storage array resources and configurations are dynamically adjusted to resolve the identified performance bottlenecks.

Dynamic performance optimization in storage arrays

A data storage node includes a plurality of compute nodes that allocate portions of local memory to a shared cache. The shared cache is configured with mirrored and non-mirrored segments. The mirrored and non-mirrored segments are separately configured with pools of data slots. Within each segment, each pool is associated with same-size data slots that differ in size relative to the data slots of other pools. The sizes of the pools in each segment are adjusted to minimize cache loss, where cache loss is the difference between the amount of shared cache used to service IOs and the minimum amount required for servicing the IOs.

Using cache loss signal as a basis to optimize hit rate and utilization through cache partitioning

A storage system is configured with pools of processor cores. Each pool corresponds uniquely to one of the supported service levels of the storage system. Processor cores within each pool run at a clock speed that is defined for the service level corresponding to the respective pool. Incoming IOs are enqueued for processing by cores of the pool corresponding to the service level of the storage group containing the storage object that is the target of the IO. Forecast IO demand and corresponding data access latency are computed for each service level. If the forecast latency of a service level includes values outside a compliance range, then cores are promoted to, or demoted from, the corresponding pool so that forecast latency stays within the compliance range, thereby reducing power consumption while promoting service level compliance.

Composable core matrix for service level compliance

One or more aspects of the present disclosure relate to autonomous write fencing for backup power. In embodiments, an input/output (IO) workload is received at a storage array. Additionally, one or more IO write requests in the IO workload are cached based on a charge level of a backup power supply of the storage array. Further, the cached IO write requests are processed.

Dataset integrity protection

A method for use in a storage system, comprising: instantiating a plurality of virtual storage devices, each virtual storage device corresponding to a different respective group of physical storage devices; assigning one or more of the plurality of virtual storage devices to a landing zone and assigning the rest of the plurality of virtual storage devices to an idle zone; instantiating a data volume, the data volume including a plurality of portions, each portion being hosted on a different respective one of the plurality of virtual storage devices; receiving an I/O request associated with the data volume, the I/O request being one of a read request or a write request; executing the I/O request.

Green disk array enclosure

Host-resident software such as a Multi-Path Input-Output (MPIO) driver generates notifications of scheduled host shutdowns and unscheduled host shutdowns. A storage node generates emergency snapshots of storage objects masked to the host server in response to receipt of an unscheduled shutdown notification. The storage node also generates emergency snapshots of storage objects masked to a host server in response to detection of a reboot sequence of the host server when a corresponding scheduled shutdown notification was not received beforehand.

Host anomaly snaps

A method, comprising: obtaining a parsed trace log that is generated by parsing a raw trace log of a storage system, the parsed trace log identifying a first sequence of tasks that are executed as part of a same input-output (I/O) operation; classifying the first sequence of tasks by using a machine learning model, the first sequence of tasks being classified into one of a first category and a second category, the first category corresponding to anomalous events, and the second category corresponding to non-anomalous events; when the first sequence of tasks is classified in the first category: identifying an anomaly code that corresponds to the first sequence of tasks and generating an alert that is associated with the anomaly code, the anomaly code being identified based on a distance between the first sequence of tasks and a second sequence of tasks that is mapped to the anomaly code.

I/O path anomaly detection

One or more aspects of the present disclosure relate to read-skew-based data theft mitigation. In embodiments, an input/output (IO) workload is received at a storage array. Further, access to a logical device can be controlled based on at least one read capacity skew forecast corresponding to the logical device.

Read skew based data theft mitigation

Autonomous power control is provided on a storage system to enable service level maximum response time controls to be achieved, while also minimizing power consumption, by enabling service level minimum response times to be specified and enforced. A workload/CPU clock speed model is created for the storage system correlating a maximum number of IOPS that the storage system can process for different CPU clock speeds for each workload type. A service level agreement specifies a maximum storage system response time. A storage system minimum response time is also specified, that is used to identify a target CPU clock speed for the workload type being provided by a host. The CPU clock speed is lowered to reduce energy consumption by the storage system toward the target CPU clock speed, and storage system performance is monitored to ensure that the storage system complies with the storage system maximum response time.

Autonomous power control based on service level compliance

Streaming machine telemetry (SMT) event counters are placed in critical code paths of software executing on a storage system. Each monitoring interval the values of the SMT counters are reported. When a critical error occurs on the storage system, a time series set of SMT counters from a set of previous monitoring intervals is labeled with the error type and used as a training example for a learning process. The learning process is trained to learn to learn recursions between time series sets of SMT counter values and labeled error types. Once trained, a checkpoint of the learning model is deployed as an inference model and used to predict the likely occurrence of errors before the errors occur. Predicted errors are logged into a proactive service request queue, and feedback related to the predicted errors are used as feedback to continue training the learning process.

Proactive service requests for forecast storage system failures

In a storage system in which processor cores are exclusively allocated to run process threads of individual emulations, the allocations of cores to emulations are dynamically reconfigured based on forecasted workload. A workload configuration model is created by testing different core allocation permutations with different workloads. The best performing permutations are stored in the model as workload configurations. The workload configurations are characterized by counts of tasks required to service the workloads. Actual task counts are monitored during normal operation and used to forecast changes in actual task counts. The forecasted task counts are compared with the task counts of the workload configurations of the model to select the best match. Allocation of cores is reconfigured to the best match workload configuration.

Autonomous core affinity

A method and apparatus for enabling dynamic selection between compression algorithms to achieve a target data reduction ratio is provided to predict, on a per-extent basis, whether a write operation targeting a piece of data on a particular extent would be preferentially compressed using either compression algorithm A or compression algorithm B. By selectively implementing compression of particular write operations using compression algorithm B, when it is predicted to achieve a greater compression benefits, it is possible to reduce the overall data reduction ratio achieved by the compression engine while economizing the amount of resources expended by the compression engine. The dynamic selection process determines a per-extent compressibility ratio, which is used in connection with a first threshold to evaluate write operations on a per-extent basis. The dynamic selection processes also determines a per-extent size threshold to select between compression algorithms for particular write operations.

Dynamic selection between compression algorithms to achieve a target data reduction ratio

Feature usage profiles of shared global memory are monitored and the profiles are used to allocate computational resources (CPU cores) between thread groups. A global memory monitor is used to monitor global memory metadata to determine the amount of space currently being consumed by a particular feature or service, and dynamically adjusts the amount of computational resources allocated to the thread group responsible for processing IO operations related to the feature or service. By exploiting an inversely proportional relationship between allocated computational resources and usage of shared global memory, it is possible to monitor feature usage profiles of shared global memory to dynamically allocate computational resources to thread groups. By looking at anomalies in usage of slots of shared global memory, and in a targeted way speeding up the processes working on such slots, it is possible to improve overall utilization of shared global memory.

Method and apparatus for dynamic CPU allocation based on feature usage profiles of shared global memory

A power controller for a data storage system automatically selects different power modes based on changing forecasts of availability of renewable energy. The power modes indicate settings for each of a plurality of hardware resources, such as processors, fans, memory, and drives. At times when the storage system can operate at less than the maximum level of performance in terms of IO latency or IOPS, power consumption of one or more of the hardware resources is reduced to a degree corresponding to availability of renewable energy. For example, the hardware resources may be operated in a relatively higher power mode when renewable energy is available than when renewable energy is unavailable. The storage system may be configured to automatically implement a low power mode when power from all sources is in limited supply, e.g., during a brownout.

Autonomous power management for sustainable storage

Growing community of inventors

Hopedale, MA, United States of America

Owen Martin