Netapp, Inc.

Solidfire, Inc.

Michael Xu

David D Wright

Charles Randall

James M Wilson

Examples described herein are generally directed to providing a fair amount of throughput to multiple clients of a storage system. According to one example, the storage system determines system metrics based upon use of the storage system by the clients. A system load value is calculated based upon the system metrics. The system load value quantifies a load on the storage system by the clients. Client loads, each representing a contribution to the use of the storage system by an individual client, are calculated and expressed as a relative degree or amount of the system load value. A target performance value for a client is calculated based on the client load for the client and a quality of service parameter value provisioned for the client to access the storage system. Based on the target performance value, latency for the client is introduced or removed.

Proportional quality of service based on client usage and system metrics

A distributed storage system monitors one or more system performance metrics and one or more client performance metrics related usage of the distributed storage system, including a read latency metric, a write latency metric, a total input/output (I/O) operations per second (IOPS) metric, a read IOPS metric, a write IOPS metric, an I/O size metric, a total bandwidth metric, a read bandwidth metric, a write bandwidth metric, a read/write ratio metric or statistical measures thereof over a period of time. When the distributed storage system is determined to be in an overload condition (e.g., when a system load value, calculated based on the performance metrics, exceeds a threshold), the distributed storage system independently throttles access to one or more components of the distributed storage system by one or more of multiple clients performing I/O operations to the distributed storage system based on their respective contribution to the overload condition.

Proportional quality of service based on client impact on an overload condition

Performance of a storage system with data distributed substantially, evenly across a cluster of storage nodes can be dynamically managed according whether quality of service (QoS) rules based on one or more performance capacities according to performance class is satisfied, wherein each QoS rule includes a boundary condition and a threshold expressing performance guarantees to a respective performance class. Determining for each QoS rule, whether a respective boundary condition satisfies a respective threshold based on measured capacity for a performance metric corresponding to the respective boundary condition; and if so, setting a first QoS parameter associated with the set of QoS rules and throttling access to the storage system by a first client associated with a first performance class based on the first QoS parameter to enforce the performance guarantees to the first performance class.

Rule-based performance class access management for storage cluster performance guarantees

A storage system monitors system performance metrics and client performance metrics related to input/output (I/O) operations. The storage system, generates a performance adjustment value for each client using a proportional-integral-derivative (PID) controller to match the corresponding respective target performance value within a time period based on feedback of the respective monitored client performance metric and the system performance metric. The storage system, throttles I/O operations for each client during the time period such that the respective client performance moves toward the corresponding respective target performance value.

Quality of service policy based load adaption

Performance of a distributed storage system with data distributed substantially, evenly across a cluster of storage devices can be dynamically managed of the distributed storage system with respect to performance guarantees to clients of the distributed storage system. Capacity of the distributed storage system in terms of one or more metrics can be determined. This measured capacity can then be compared with allocations of the metric(s) to clients of the distributed storage system. The allocations are determined based on quality of service parameters specified for the clients. The quality of service parameters at least include a maximum value and a minimum value for each of the one or more metrics, and can also include burst credits allocated to the clients. Access to the distributed storage system by the clients can be throttled to ensure the performance guarantees corresponding to the quality of service parameters are fulfilled.

Managing client access for storage cluster performance guarantees

A distributed storage system can maintain calculated load values of services throughout the system. The load values represent different types of load on the system. When a load value corresponding to a client changes, the load value is evaluated against thresholds to determine whether a different quality of service (QoS) policy is to be selected and implemented. If so, a function/formula indicated in the selected policy is used to calculate a scaling factor, which is then used to calculate a target performance value based on QoS parameters of the client. The system then manages access by the client according to the target performance value.

Disclosed are systems, computer-readable mediums, and methods for determining client metrics of a volume in a storage system for a first client of a plurality of clients. The storage system stores data from the plurality of clients. System metrics of a cluster in the storage system are determined based upon use of the storage system by the plurality of clients. A load value of the storage system is determined based upon the system metrics and the client metrics. The load value is determined to be above a predefined threshold. A target performance value is calculated based upon the load value, a minimum quality of service value, and a maximum quality of service value. Performance of the storage system is adjusted for the client based upon the target performance value and the determining the load value is above the predefined threshold.

Disclosed are systems, computer-readable mediums, and methods for receiving client quality of service values associated with a first client. Client metrics of a volume in a storage system are determined for the first client. The client metrics reflect usage of the storage system by the first client. A client load value based is determined upon the client metrics. A quality of service management policy for the first client is determined from a plurality of quality of service management policies based upon the client load value. Each quality of service management policy comprises a formula based on a quality of service parameter. A client target performance value is determined based upon the formula of the quality of service management policy and the received client quality of service values. Performance of the storage system is adjusted for the first client based upon the client target performance value.

Quality of service policy sets

Updating a second cluster server that backs up a first cluster server includes retrieving a first metadata file from a first cluster server. The first metadata file includes a first ordered list of block identifiers for data blocks stored on a first plurality of block servers. The updating also includes retrieving a second metadata file from a second cluster server. The second metadata file includes a second ordered list of block identifiers for data blocks stored on a second plurality of block servers. The updating also includes comparing the first metadata file to the second metadata file to determine a difference list. The difference list includes block identifiers from the first ordered list that differ from block identifiers of the second ordered list. The updating also includes sending, to the first cluster server, a request for data blocks associated with the block identifiers from the difference list.

Parallel optimized remote synchronization of active block storage

Disclosed are systems, computer-readable mediums, and methods for transforming data in a file system. As part of a recycling process, a determination is made that transformations should be attempted. A data block is determined to be in use by at least one user of the storage system. If a transformation should be attempted on the data block is determined. Parameters associated with the performance of the file system can be used in this determination. A type of transformation to be done is determined. The data block is transformed based upon the selected transformation. The transformed data block is written to the storage system. As part of the recycling process, the transformation requires no additional input/output requests.

Data transformation during recycling

In one embodiment, a method includes determining a previous client performance value in terms of a performance metric for a volume in a storage system. The previous client performance value is related to previous access for a client to the volume. Also, the storage system is storing data for a plurality of volumes where data for each of the plurality of volumes is striped substantially evenly across drives of the storage system. The method applies criteria to the previous performance value to determine a target performance value. Performance of the client with respect to access to the volume is regulated in terms of the performance metric based on the target performance value.

Growing community of inventors

Boulder, CO, United States of America

Michael Xu