Storage system and method for controlling storage system

The storage system dynamically adjusts resources based on load calculations to maintain performance stability and comply with user demands in fluctuating public cloud environments.

JP2024162866AInactive Publication Date: 2024-11-21HITACHI VANTARA LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2023078812
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-11
Publication Date
2024-11-21
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing storage cloud systems struggle to efficiently increase or decrease storage resources without causing performance degradation, particularly in public cloud environments where user demands fluctuate.

Method used

A storage system and method that calculates load values, determines resource additions or reductions based on predetermined thresholds, and performs expansion or contraction processes to maintain stable performance by dynamically adjusting CPU and memory resources and node numbers.

Benefits of technology

The system stabilizes storage performance by flexibly adding or removing resources in response to load changes, ensuring efficient operation and adherence to contracted capacities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024162866000001_ABST
    Figure 2024162866000001_ABST
Patent Text Reader

Abstract

To operate a storage having a cluster configuration while stabilizing performance thereof.SOLUTION: A storage system 20 calculates a load value of a storage in which a predetermined cluster configuration is set, determines whether the calculated load value exceeds a predetermined value, adds a predetermined resource to the storage when the calculated load value exceeds the predetermined value, calculates a predicted value of a load of the storage when the resource is reduced after the resource is added, determines whether the calculated predicted value is lower than a predetermined value, and reduces the resource of the storage when the calculated predicted value is lower than the predetermined value.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a storage system and a control method for a storage system. [Background technology]

[0002] Technologies are being developed for building storage with a clustering configuration on the cloud (cloud storage).

[0003] For example, Patent Document 1 discloses a volume placement control method as a technique for controlling volume placement in a storage cluster that provides a virtual storage system, in which when a new node (a replaced or newly added storage node) is added to a storage cluster, the storage system compares the specifications of the new node with the specifications of at least one existing node other than the new node, and if the specifications of the new node are higher than the specifications of the existing nodes, the storage system determines that if any of the existing nodes has a first volume (a volume associated with a priority higher than the first priority), the migration destination of the first volume is the new node. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2022-110245 Summary of the Invention [Problem to be solved by the invention]

[0005] Here, if a user of a storage cloud wants to use the storage cloud comfortably by using a public cloud or the like, it is preferable to be able to increase or decrease storage resources according to the user's circumstances, especially to increase or decrease resources to ensure temporary performance, even in a situation where the user is writing data.However, the current situation is that technology that allows such an increase or decrease to be performed stably while suppressing performance degradation (so that I / O load can be reduced) has not been sufficiently developed.

[0006] The present invention has been made in consideration of the above background, and its object is to provide a storage system and a control method for a storage system that are capable of operating storage having a cluster configuration while stabilizing its performance. [Means for solving the problem]

[0007] One of the present inventions for solving the above problem is a storage system comprising an arithmetic device that executes the following operations: calculates a load value of a storage having a specified cluster configuration set, judges whether the calculated load value exceeds a specified value, and, if the calculated load value exceeds the specified value, performs an expansion process of adding specified resources to the storage; and calculates a predicted value of the load of the storage when the resources are reduced after the resources are added, judges whether the calculated predicted value is below a specified value, and, if the calculated predicted value is below the specified value, performs a reduction process of reducing the resources of the storage. Effect of the Invention

[0008] According to the present invention, it is possible to operate a storage having a cluster configuration while stabilizing its performance. Configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0009] [Figure 1]FIG. 1 is a diagram illustrating an example of a configuration of an information processing system according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram illustrating main functions of the service portal platform. [Diagram 3] FIG. 13 illustrates an example of a host management table. [Figure 4] FIG. 13 illustrates an example of a threshold management table. [Diagram 5] FIG. 2 is a diagram illustrating an example of a hardware configuration of each information processing device included in the information processing system. [Figure 6] FIG. 2 is a diagram illustrating an overview of a storage control process performed in the information processing system. [Figure 7] FIG. 11 is a flow diagram illustrating an example of a configuration information acquisition process. [Figure 8] FIG. 13 is a diagram illustrating an example of a storage node management table. [Figure 9] FIG. 11 is a flow diagram illustrating an example of a measurement process. [Figure 10] 13 is a diagram illustrating an example of an IOPS measurement result management table. [Figure 11] FIG. 11 is a flow diagram illustrating details of the expansion / reduction process. [Figure 12] FIG. 11 is a flow diagram illustrating details of the expansion determination process. [Figure 13] FIG. 11 is a flow diagram illustrating details of a resource addition determination process. [Figure 14] A flow chart explaining details of the reduction determination process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0011] 1 is a diagram showing an example of the configuration of an information processing system 1 according to this embodiment. The information processing system 1 includes one or more user terminals 10 used by each user, and a storage system 20 that provides the user with a volume 42 generated based on a storage pool 41, which will be described later. The storage system 20 and the user terminal 10 are communicatively connected to each other via a wireless or wired communication network 5, such as the Internet, a LAN (Local Area Network), a WAN, or a dedicated line.

[0012] The storage system 20 is an information processing system (so-called public cloud) provided to each user by a predetermined business operator. The storage system 20 includes a user cloud system 30 accessed from a user terminal 10 and a storage cloud system 40 including a storage pool 41.

[0013] The user cloud system 30 includes a service portal operation instance 31 and a host 32 provided for each user (user terminal 10).

[0014] The service portal operation instance 31 transmits a volume control instruction to the service portal platform 43 (described later) to instruct the capacity of the volume 42 to be changed.

[0015] The host 32 can access the volume 42 based on input from the user terminal 10 or the like, and can create and store data.

[0016] In this embodiment, the service portal operation instance 31 and the host 32 are virtual servers constructed based on one or more (physical) information processing devices (i.e., equipped with virtual CPUs 33, 35 and memories 34, 36), but may also be configured as physical servers.

[0017] Next, the storage cloud system 40 includes the following systems: a service portal platform 43 , a controller node 44 , a storage node group 45 , and a storage pool 41 .

[0018] The storage pool 41 is a virtual storage constructed based on a storage node group 45 described later. The storage pool 41 is constructed as storage having a predetermined cluster configuration (having redundancy). The storage pool 41 is constructed by allocating a plurality of storage nodes 46 of the storage node group 45. For example, the storage pool 41 has a cluster configuration with a predetermined number of data storages (data disks) and a predetermined number of parity storages (parity disks), and each data is distributed and stored in each cluster.

[0019] The volume 42 is made up of one or more storage areas (a so-called storage cloud) reconfigured based on the storage pool 41. The user terminal 10 can access the volume 42 via the host 32.

[0020] The service portal platform 43 executes capacity change processing for the volume 42. That is, based on the volume change instruction received from the service portal operation instance 31, the service portal platform 43 instructs the controller node 44 to change the capacity of the volume 42 to the capacity indicated by the volume change instruction, and the controller node 44 executes this. On the other hand, if the capacity indicated by the volume change instruction exceeds a predetermined capacity (contracted capacity) previously associated with the host 32 (user), the service portal platform 43 rejects the instruction. The contracted capacity can be acquired, for example, from the host management table 100 described later.

[0021] Furthermore, the service portal platform 43 instructs the controller node 44 to change the resources of the storage pool 41. Based on the instruction, the controller node 44 controls the storage node group 45 and executes the change of the resources of the storage pool 41. In this embodiment, the change of resources includes an increase or decrease (addition and reduction) in the number of storage nodes 46 assigned to the storage pool 41, and an increase or decrease in the performance of a CPU 47 and a memory 48 of the storage node 46, which will be described later, but may also include an increase or decrease in other resources. Note that when the resources are changed, the capacity of the storage pool 41 provided to the user (the capacity provided to the user as the volume 42) is not changed.

[0022] The controller node 44 controls the storage node group 45 based on instructions received from the service portal platform 43 .

[0023] In this embodiment, the service portal platform 43 and the controller node 44 are virtual servers constructed based on one or more (physical) information processing devices (i.e., having virtual CPUs 50, 53, memories 51, 54, and storage devices 53, 55), but may also be configured as physical servers.

[0024] The storage node group 45 realizes a public cloud as described above, and is composed of a plurality of storage nodes 46 connected to each other so as to be able to communicate with each other. The storage node 46 is storage, which is a resource allocated to the storage pool 41 used by the user. In this embodiment, the storage node 46 is a virtual server constructed based on one or a plurality of (physical) information processing devices (i.e., includes a virtual CPU 47, memory 48, and storage device 49), but may be configured as a physical server.

[0025] Next, the functions of the service portal platform 43 will be described. FIG. 2 is a diagram for explaining the main functions of the service portal platform 43. As shown in FIG.

[0026] The service portal platform 43 stores a host management table 100 that stores information about clusters (hereinafter referred to as storage clusters) of the storage pool 41 assigned to each host 32, a storage node management table 200 that stores the configuration of the storage node 46, an IOPS measurement result management table 300 that accumulates the measurement results of the load of each cluster of the storage pool 41, and a threshold management table 400 that stores information on threshold patterns (hereinafter referred to as threshold rules) for determining whether or not to change the configuration of the storage cluster of the storage pool 41 based on the load measurement results. Note that the service portal platform 43 is assumed to store in advance threshold rules to be used (hereinafter referred to as usage rules) from among the threshold rules stored in the threshold management table 400.

[0027] Next, the service portal board 43 has the following functional units: a measurement unit 61, an expansion determination unit 62, and a reduction determination unit 63.

[0028] The measurement unit 61 calculates a load value of the storage pool 41 (storage nodes 46 having a predetermined cluster configuration). In this embodiment, the load value is a value (a value between 0% and 100%) obtained by dividing the measured number of times (IOPS: Input / Output Per Second) that data can be accessed (data can be read or written) to the storage node 46 per second by the maximum number of times (the number of times when maximum performance is exerted), but other indices may be used.

[0029] Specifically, first, the measurement unit 61 calculates the performance value of data access of the storage node 46 for a specified period of time, and then calculates the accumulated value for the specified period of time of the difference between the calculated performance value and a specified threshold value (hereinafter referred to as the additional IO load threshold value).

[0030] In this embodiment, the measurement unit 61 calculates an accumulation value corresponding to a performance value related to reading data (hereinafter referred to as a read excess area) and an accumulation value corresponding to a performance value related to writing data (hereinafter referred to as a write excess area).

[0031] Then, the measurement unit 61 calculates, as the load value, the ratio of the calculated excess accumulated value (read excess area, write excess area) to the accumulated value (reference accumulated value) of the performance value during the period when the performance value was a specified value (in this embodiment, the performance value was 100%).

[0032] In this embodiment, the measurement unit 61 calculates, with respect to the reference accumulation value, a reference accumulation value corresponding to a performance value related to reading data (hereinafter referred to as a read total area) and a reference accumulation value corresponding to a performance value related to writing data (hereinafter referred to as a write total area).Then, with respect to the above ratio (load value), the measurement unit 61 calculates, with respect to the above ratio (load value), a ratio corresponding to a performance value related to reading data (hereinafter referred to as a read excess ratio) and a ratio corresponding to a performance value related to writing data (hereinafter referred to as a write excess ratio).

[0033] Furthermore, the measurement unit 61 calculates a performance value of the storage pool 41 (storage node 46) when the expansion determination unit 62 adds resources to the storage pool 41 as described below and then reduces the resources. In this embodiment, the performance value is the number of times data access (data reading or writing) to the storage is possible per second (IOPS), but other indices may be used.

[0034] Specifically, the measurement unit 61 calculates a predicted value of the load related to data reading and data writing when the expansion determination unit 62 increases processing performance or adds nodes as described below, and then decreases processing performance or decreases the number of nodes.

[0035] Next, the expansion determination unit 62 determines whether the load value calculated by the measurement unit 61 exceeds a predetermined value, and if the calculated load value exceeds the predetermined value, adds a specified resource to the storage pool 41.

[0036] Specifically, if the load value (read excess ratio) related to data reading calculated by the measurement unit 61 exceeds a predetermined value (hereinafter referred to as the expansion excess threshold), the expansion determination unit 62 increases the processing performance of the storage node 46 corresponding to the storage pool 41 (for example, increases the processing performance of the CPU 47 and memory 48), and if the load value (write excess ratio) related to data writing calculated by the measurement unit 61 exceeds a predetermined value (expansion excess threshold), the expansion determination unit 62 increases the storage node 46 to form a cluster of the storage pool 41 (for example, adds a new storage node 46 to the storage pool 41), thereby forming a new cluster.

[0037] Next, the reduction determination unit 63 determines whether the predicted value calculated by the measurement unit 61 is below a predetermined value (hereinafter referred to as the reduction IO load threshold), and if the calculated predicted value is below the predetermined value, reduces the resources of the storage pool 41 (storage node 46).

[0038] Specifically, if the predicted values ​​for data reading and data writing calculated by the measurement unit 61 are below a predetermined value, the reduction determination unit 63 reduces the processing performance (e.g., reduces the processing performance of the CPU 47 and memory 48) or reduces the number of nodes (e.g., removes the storage node 46 from the allocation to the storage pool 41).

[0039] (Host Management Table) 3 is a diagram showing an example of the host management table 100. The host management table 100 has the following data items: host ID 101 in which the ID of the host 32 is set, storage cluster ID 102 in which the identifier of the storage cluster set for the host 32 is set, and contracted capacity 103 in which the capacity allocated to the host 32 is set. The host management table 100 may be set by an administrator, for example, or the storage cloud system 40 may automatically acquire information and set it in the host management table 100.

[0040] (Threshold management table) 4 is a diagram showing an example of the threshold management table 400. The threshold management table 400 has the following data items: a threshold ID 401 in which the ID of a threshold rule is set, an expansion IO load 402 in which an expansion IO load threshold in the threshold rule is set, an excess expansion threshold 403 in which an excess expansion threshold in the threshold rule is set, a reduction IO load threshold 404 in which a reduction IO load threshold in the threshold rule is set, a collection interval 405 in which a collection time interval (data collection interval) of data (performance value) in a period (measurement period) in which data related to the load of data access is collected is set, and a collection period 406 in which a period in which data (performance value) related to the load of data access is collected is set.

[0041] The threshold management table 400 is set in advance by, for example, an administrator or a user. For example, the service portal platform 43 displays a specified screen and accepts input of each value from the administrator or user. For example, the collection period is the time period from 9:00 to 17:00 on weekdays in a certain week, and the data collection interval is every minute. Also, for example, the reduction IO load threshold and the expansion IO load threshold are set to any value between 40% and 100% (for example, 70%).

[0042] Next, FIG. 5 is a diagram showing an example of a hardware configuration of each (physical) information processing device constituting the information processing system 1. Each information processing device includes a calculation device 91 such as a CPU (Central Processing Unit), a memory 92 such as a RAM (Random Access Memory) or a ROM (Read Only Memory), an external storage device 93 such as a HDD (Hard Disk Drive) or an SSD (Solid State Drive), an input device 94 such as a keyboard, a mouse, or a touch panel, an output device 95 such as a display or a touch panel, and a communication device 96 configured with a NIC (Network Interface Card), a wireless communication module, a USB (Universal Serial Interface) module, a serial communication module, or the like. The functions of the functional parts of each information processing device (virtual server) in the information processing system 1 described above are realized by the calculation device 91 reading a program from the memory 92 or the external storage device 93. In addition, each program can be distributed by recording it on a portable or fixed recording medium, for example. Next, the processes performed in the information processing system 1 will be described.

[0043] FIG. 6 is a diagram illustrating an overview of the storage control process performed in the information processing system 1. As shown in FIG.

[0044] First, the storage cloud system 40 executes a configuration information acquisition process s1 to acquire configuration information of the storage node 46. The configuration information acquisition process s1 is started, for example, when a predetermined input is made to the service portal platform 43.

[0045] After executing the configuration information acquisition process s1, the storage cloud system 40 executes a measurement process s3 to measure the current load of each cluster (storage node 46) of the storage pool 41. The measurement process s3 is executed repeatedly at a predetermined timing (for example, at a predetermined time, at a predetermined time interval), for example.

[0046] After executing the measurement process s3 at least once, the storage cloud system 40 judges whether to change the resources of the storage pool 41 (for example, temporary addition or subsequent removal of storage nodes 46) and executes the expansion / reduction process s5 to change the resources according to the judgment result. The expansion / reduction process s5 is repeatedly executed at a predetermined timing (for example, a predetermined time, a predetermined time interval) after executing the measurement process s3. Each process will be described in detail below.

[0047] <Configuration information acquisition process> 7 is a flow diagram illustrating an example of the configuration information acquisition process s1. The service portal platform 43 acquires information on the configuration and storage cluster from each storage node 46 of the storage node group 45 (s11), and records the acquired information in the storage node management table 200 (s12).

[0048] (Storage node management table) 8 is a diagram showing an example of a storage node management table 200. The storage node management table 200 has the following data items: storage cluster ID 201 in which the ID of a storage cluster is set, storage node ID 202 in which the storage nodes 46 constituting the storage cluster are set, cluster configuration 203 in which the configuration of the storage cluster (for example, the number of data disks and the number of parity disks constituting the storage cluster) is set, and capacity 204 in which the capacity of the storage node 46 is set.

[0049] <Measurement processing> Next, FIG. 9 is a flow diagram illustrating an example of the measurement process s3.

[0050] The measurement unit 61 determines whether the current date and time is the measurement start date and time (s31). For example, the measurement unit 61 makes this determination based on the date and time of a preset schedule or the contents of the collection period 406 of a record related to the usage rule in the threshold management table 400.

[0051] If the current date and time is the measurement start date and time (s31: YES), the measurement unit 61 executes the process of s32. If the current date and time is not the measurement start date and time (s31: NO), the measurement process s3 ends.

[0052] In s32, the measurement unit 61 measures data access for each storage cluster (storage node 46) during a measurement period and calculates a performance value.

[0053] For example, the measurement unit 61 calculates the performance value by performing data access (reading and writing data) to each storage node 46 at each time of the data collection interval indicated by the collection interval 405 of the record related to the usage rule during the measurement period indicated by the collection period 406 of the record in the threshold management table 400.

[0054] Then, the measurement unit 61 performs a predetermined processing (such as normalization) on the performance values ​​calculated in s32 (s33).

[0055] Specifically, first, the measurement unit 61 calculates "data reading performance value-expansion IO load threshold" as an excess expansion difference at each time of the measurement period for the reading of the above data. Meanwhile, the measurement unit 61 calculates "data reading performance value-expansion IO load threshold" as an excess reduction difference for the reading of data at each data collection period of the measurement period.

[0056] Then, the measurement unit 61 calculates the total value of all the excess expansion differences during the measurement period. That is, the measurement unit 61 calculates the lead excess area for each data collection period with respect to the excess expansion difference.

[0057] Furthermore, the measurement unit 61 calculates the total value of the performance value during the period when the excess expansion difference was 0. In other words, the measurement unit 61 calculates the total lead area for the period when the load was 100%.

[0058] Then, the measuring unit 61 calculates the lead excess ratio (i.e., load value) by dividing the lead excess area by the total lead area.

[0059] On the other hand, the measurement unit 61 performs the same process for data writing. That is, for the above data writing, the measurement unit 61 calculates "data writing performance value-expansion IO load threshold" as an excess expansion difference at each time of the measurement period. On the other hand, the measurement unit 61 calculates "data writing performance value-reduction IO load threshold" as an excess reduction difference at each data collection period of the measurement period.

[0060] Then, the measurement unit 61 calculates the total value of all the excess expansion differences during the measurement period. That is, the measurement unit 61 calculates the write excess area for each data collection period with respect to the excess expansion difference.

[0061] Furthermore, the measurement unit 61 calculates the total value of the performance value during the period when the excess expansion difference was 0. In other words, the measurement unit 61 calculates the entire light area during the period when the load was 100%.

[0062] Then, the measurement unit 61 calculates the write excess ratio (that is, the load value) by dividing the write excess area by the total write area.

[0063] Next, the measurement unit 61 updates the IOPS measurement result management table 300 with the values ​​calculated in s33 (s34).

[0064] (IOPS measurement result management table) 10 is a diagram showing an example of an IOPS measurement result management table 300. The IOPS measurement result management table 300 has data items of storage cluster ID 301 in which the ID of a storage cluster is set, measurement start date and time 302 in which the date and time (measurement start date and time) when the measurement of the load of data access related to the storage cluster is started is set, threshold rule 303 in which a specified threshold rule (usage rule) is set, read IO excess ratio 304 in which the read excess ratio in the measurement period started from the measurement start date and time is set, write IO excess ratio 305 in which the write excess ratio in the measurement period started from the measurement start date and time is set, read IO continuous time 306 in which the time (read excess continuous time) during which the load value related to reading data continuously exceeds the expansion excess threshold in the measurement period started from the measurement start date and time is set, write IO continuous time 307 in which the time (write excess continuous time) during which the load value related to writing data continuously exceeds the expansion excess threshold in the measurement period started from the measurement start date and time is set, and total IO load 308 in which a total IO load value is set.

[0065] In the process of s34, for example, the measurement unit 61 sets the calculated read excess ratio to the read IO excess ratio 304 in the IOPS measurement result management table 300. Also, the measurement unit 61 sets the calculated write excess ratio to the write IO excess ratio 305 in the IOPS measurement result management table 300.

[0066] The measurement unit 61 also identifies the length of time during the measurement period during which the excess expansion difference (or reduction increase difference) for data reading continuously exceeds 0, and sets the identified length as the read IO continuous time 306. The measurement unit 61 also identifies the length of time during the measurement period during which the excess expansion difference (or reduction increase difference) for data writing continuously exceeds 0, and sets the identified length as the write IO continuous time 307. The measurement unit 61 also sets the average performance value (total IO load value) of data access (data reading and data writing) during the measurement period as the total IO load 308.

[0067] In the above, the measuring unit 61 calculates the load value from the values ​​(excess read ratio, excess write ratio) normalized by the total read area and the total rewrite area, but normalization may be omitted. Also, the performance value may not be normalized and the IOPS value itself may be used.

[0068] <Increase / decrease processing> FIG. 11 is a flow diagram illustrating the details of the expansion / reduction process s5.

[0069] First, the expansion determination unit 62 determines whether or not to add resources to the storage pool 41 based on the IOPS measurement result management table 300, and executes an expansion determination process s51 for adding resources when it has determined whether or not to add resources. Note that the expansion determination process s51 may be executed repeatedly.

[0070] If a resource is added in the expansion judgment process s51 (s52: YES), the expansion judgment unit 62 executes the reduction judgment process s52, and if a resource is not added in the expansion judgment process s51 (s52: NO), the expansion judgment unit 62 repeats the expansion judgment process s51.

[0071] In s53, the reduction determination unit 63 determines whether or not to reduce the resources that were (temporarily) added in the expansion determination process s51 for the storage pool 41 based on the IOPS measurement result management table 300, and executes a reduction determination process s53 for reducing the resources if it is determined that the resources should be reduced. The reduction determination process s53 may be executed repeatedly (however, the number of times does not exceed the number of times the expansion determination process s51 is executed).

[0072] If a resource reduction is performed in the reduction judgment process s53 (s54: YES), the reduction judgment unit 63 repeats the expansion judgment process s51, and if a resource reduction is not performed in the reduction judgment process s53 (s54: NO), the reduction judgment unit 63 repeats the reduction judgment process s53.

[0073] <Expansion determination process> FIG. 12 is a flow diagram illustrating the details of the expansion determination process s52.

[0074] The expansion determination unit 62 determines whether or not a failure (hardware failure, or a failure in the CPU 47 or the memory 48, etc.) is currently occurring in any of the storage nodes 46 (s71). For example, the expansion determination unit 62 calls a predetermined monitoring tool provided in advance in the storage cloud system 40 and acquires failure information to determine the presence or absence of a failure.

[0075] If a failure is currently occurring in any of the storage nodes 46 (S71: YES), the expansion determination unit 62 records that a storage cluster is to be added as a resource addition and executes processing of s73; if no failure is currently occurring in any of the storage nodes 46 (S71: NO), the expansion determination unit 62 executes processing of s80.

[0076] In s80, the addition determination unit 62 executes a resource addition determination process s80 for determining whether or not to add resources to the storage pool 41.

[0077] If the addition determination unit 62 determines that resources are to be added (s72: YES), it executes the process of s73, and if it determines that resources are not to be added (s72: NO), the addition determination unit 62 repeats the process of s71.

[0078] In s73, the expansion determination unit 62 instructs the controller node 44 to add resources based on the contents stored in s71 or the contents stored as a processing result of the resource addition determination process s80 (described later). Then, the expansion determination unit 62 waits to receive a processing completion notification from the controller node 44 (s74: NO).

[0079] On the other hand, the controller node 44 performs processing based on the instruction received from the expansion determination unit 62. For example, the controller node 44 newly allocates a predetermined number of storage nodes 46 to the storage pool 41 so as to maintain the redundancy configuration of the current storage pool 41. For example, if the current cluster configuration is a cluster consisting of four data disks and two parity disks, the controller node 44 newly allocates six storage nodes 46 consisting of four data disks and two parity disks. Note that the controller node 44 does not change the capacity of the storage pool 41 itself. Note that information on the redundancy of the storage pool is acquired, for example, from the storage node management table 200.

[0080] Also, for example, the controller node 44 increases the performance of the CPU 47 and memory 48 of the storage node 46 currently assigned to the storage pool 41 by a predetermined amount (for example, doubles the CPU clock speed and memory capacity).

[0081] When the addition determination unit 62 receives the completion notification from the controller node 44 (s74: YES), it executes a resource addition determination process s80.

[0082] Thereafter, if the resource addition determination process s80 determines that resources should be added (s75: YES), the expansion determination unit 62 repeats the process of s73; if the resource addition determination process s80 determines that resources should not be added (s75: NO), the expansion determination process s52 ends.

[0083] <Resource addition judgment process> FIG. 13 is a flow diagram illustrating the details of the resource addition determination process s80.

[0084] The expansion determination unit 62 determines whether the load on the storage cluster with respect to data reading is high (s81). Specifically, the expansion determination unit 62 refers to the latest record of each storage cluster in the IOPS measurement result management table 300, and determines whether the read IO excess ratio of each record exceeds the threshold indicated by the expansion excess threshold 403 of the record related to the usage rule in the threshold management table 400.

[0085] If the load on the storage cluster related to data reading is high (s81: YES), the expansion determination unit 62 executes the processing of s82, and if the load on the storage cluster related to data reading is not high (s81: NO), the expansion determination unit 62 determines not to add resources and the resource addition determination processing s80 ends.

[0086] In s82, the expansion determination unit 62 determines whether the load on the storage cluster with respect to data writing is high. Specifically, the expansion determination unit 62 refers to the latest record of each storage cluster in the IOPS measurement result management table 300, and determines whether the write IO excess ratio of each record exceeds the threshold indicated by the expansion excess threshold 403 of the record related to the usage rule in the threshold management table 400.

[0087] If the load on the storage cluster regarding data writing is high (s82: YES), the expansion judgment unit 62 executes processing of s83, and if the load on the storage cluster regarding data writing is not high (s82: NO), the expansion judgment unit 62 executes processing of s84.

[0088] In s83, the addition determination unit 62 stores the information that a storage cluster (storage node 46) is to be added as a resource addition. This ends the resource addition determination process s80.

[0089] In s84, the expansion determination unit 62 stores the information that the processing performance (CPU and memory performance) will be increased by adding resources. This ends the resource addition determination process s80.

[0090] <Reduction judgment process> FIG. 14 is a flow diagram for explaining the details of the reduction determination process s53.

[0091] The reduction determination unit 63 acquires information on the configuration of each storage node 46 (contents of the cluster configuration 203) from the storage node management table 200 (s91).

[0092] Furthermore, the reduction determination unit 63 acquires information on the load value related to each storage cluster (for example, the contents of the latest record of each storage cluster) from the IOPS measurement result management table 300 (s92).

[0093] The reduction determination unit 63 repeats the following processing of s94 to s97 the number of times n that resources were added in the expansion determination processing s52 (s93, s100).

[0094] That is, the reduction determination unit 63 determines whether or not a storage cluster (storage node 46) has been added in the expansion determination process s52 (s94).

[0095] If a storage cluster has been added (s94: YES), the reduction determination unit 63 stores the fact that a storage cluster will be reduced as a resource reduction (s95), and then performs the processing of s97. If a storage cluster has not been added (s94: NO), the reduction determination unit 63 stores the fact that a CPU and memory performance will be reduced as a resource reduction (s96), and then performs the processing of s97.

[0096] In s97, the reduction determination unit 63 determines whether or not to actually perform a reduction in resources corresponding to one resource addition in the expansion determination process s52.

[0097] That is, the reduction determination unit 63 determines whether or not the load on data access (data reading and writing) of the storage cluster is low when resources are reduced. Specifically, the expansion determination unit 62 refers to the latest record of each storage class in the IOPS measurement result management table 300, and determines whether or not the value obtained by dividing the read total IO load 308 of each record by n (predicted value of performance value after resource reduction) is below the threshold indicated by the reduction IO load threshold 404 of the record related to the usage rule in the threshold management table 400. Note that the determination method described here is only an example, and any method for predicting whether or not the performance value (or load value) will be equal to or greater than a certain value when resources are reduced can be adopted.

[0098] If the resources are to be reduced (s97: YES), the reduction determination unit 63 executes the process of s98, and if the resources are not to be reduced (s97: NO), the reduction determination unit 63 executes the process of s100.

[0099] In s98, if the reduction determination unit 63 has stored information indicating that a storage cluster is to be reduced, it first migrates data (target data) stored in the storage nodes 46 related to the storage cluster to be reduced, among the data of the storage nodes 46 currently assigned to the storage pool 41, to storage nodes 46 (target nodes) related to storage clusters that are not to be reduced. At this time, the reduction determination unit 63 evenly distributes and migrates (rebalances) the target data to each target node.

[0100] Then, the reduction determination unit 63 transmits an instruction to reduce the resources stored in s95 or s96 to the controller node 44 together with the configuration information acquired in s91 (s99).

[0101] When the controller node 44 receives the instruction, it performs the process of reducing resources based on the instruction.

[0102] For example, when the reduction determination unit 63 stores information indicating that a storage cluster is to be reduced, the reduction determination unit 63 reduces the storage cluster (storage node 46) (for example, reduces the storage cluster corresponding to the expansion performed in the expansion determination process s52). Also, when the reduction determination unit 63 stores information indicating that the performance of the CPU and memory is to be reduced, the reduction determination unit 63 reduces the performance of the CPU 47 and memory 48 of the storage node 46 (for example, reduces the performance by an amount equal to the increase in performance performed in the expansion determination process s52). At this time, the controller node 44 does not change the capacity of the storage pool 41 itself. This ends the reduction determination process s53.

[0103] As described above, the storage cloud system 40 of this embodiment calculates a load value for storage (storage pool 41 to storage node 46) having a predetermined cluster configuration, and if the calculated load value exceeds a predetermined value, adds (adds) the predetermined resources that constitute the cluster to the storage pool 41. Furthermore, after adding the resources, it calculates a predicted load value for the storage when the resources are reduced, and if the calculated predicted value is below the predetermined value, reduces (decreases) the resources of the storage pool 41 (storage node 46).

[0104] In this way, by adding resources when the load on the storage pool 41 having a cluster configuration is heavy, and by reducing the resources when the load decreases and processing stabilizes after adding resources, it is possible to maintain the performance of the storage pool 41 with resources that are not excessive, even if the load on the storage pool 41 temporarily increases.

[0105] As described above, according to the storage cloud system 40 of this embodiment, it is possible to operate storage having a cluster configuration while stabilizing its performance.

[0106] Furthermore, in the storage cloud system 40 of this embodiment, when the load value of a storage pool 41 having a predetermined processing performance (CPU and memory) and a predetermined cluster configuration is set exceeds a predetermined value, the processing performance is increased or the number of storage nodes 46 constituting the cluster of the storage pool 41 is increased or the number of nodes is increased, and when the predicted load value for the storage in the case where the processing performance is decreased or the number of nodes is decreased after the processing performance is increased or the number of nodes is decreased, the processing performance of the storage is decreased or the number of storage nodes is decreased.

[0107] In this way, by changing the processing performance or increasing or decreasing the number of storage nodes 46 in response to an increase or decrease in the load on the storage pool 41, flexible resource changes can be made according to the magnitude of the increase or decrease in load, and operation can be performed while stabilizing performance.

[0108] In addition, the storage cloud system 40 of this embodiment increases processing performance when the load value related to data reading from the storage exceeds a predetermined value, and increases the storage nodes 46 that constitute the storage cluster when the load value related to data writing to the storage exceeds a predetermined value, and thereafter, when the predicted values ​​related to data reading and data writing in the case where the processing performance is decreased or the number of nodes is decreased are below a predetermined value, the processing performance is decreased or the number of storage nodes 46 is decreased.

[0109] In this way, when the data read load is large, only the processing capacity is increased, whereas when the data write load is large, the storage nodes 46 are increased, thereby making it possible to add the appropriate type of resource according to the severity of the load. In addition, when it comes to reducing resources, the decision to reduce resources is made based on the data read and write loads. This makes it possible to further stabilize the storage performance.

[0110] Furthermore, when the storage cloud system 40 of this embodiment reduces the number of storage nodes 46, the data stored in the storage pool 41 to which the storage nodes 46 have been added is evenly distributed and stored in each storage node 46 of the storage pool 41 after the storage nodes 46 have been reduced.

[0111] In this way, by writing data to each node after the removal while rebalancing the data, it is possible to ensure the redundancy of the storage pool 41 after the removal without compromising the data.

[0112] In addition, when adding resources, the storage cloud system 40 of this embodiment calculates a performance value of data access of the storage pool 41 (storage node 46) during a measurement period, calculates the difference between the calculated performance value and a predetermined threshold (additional IO load threshold) based on the accumulated values ​​(read excess area, write excess area) during the measurement period as a load value, and determines whether the calculated load value exceeds the predetermined threshold.

[0113] This prevents unstable expansion, such as the need to add resources due to only a temporary increase in load, and stabilizes storage performance.

[0114] Furthermore, when adding resources, the storage cloud system 40 of this embodiment calculates the load value as the ratio (read excess ratio, write excess ratio) of the calculated accumulated value (read excess area, write excess area) to the accumulated value (total read area, total write area) of the performance value during the period when the performance value was a specified value (100%), and determines whether the calculated ratio exceeds the specified ratio.

[0115] By carrying out such normalization processing, it is possible to stably determine whether or not to add resources, and to stabilize the performance of the storage.

[0116] In the storage cloud system 40 of this embodiment, the storage nodes 46 are configured as information processing devices in a plurality of information processing devices configured to be able to communicate with each other, and the storage pool 41 is configured as a cluster by the plurality of storage nodes 46.

[0117] In this way, the storage node 46 is configured as an information processing device in a cloud (for example, a public cloud), and the storage pool 41 is configured from the storage nodes 46. The storage cloud system 40 of this embodiment can stabilize the performance of the cloud storage even in such a hybrid cloud configuration.

[0118] Furthermore, when the storage cloud system 40 of this embodiment receives an instruction to change the capacity of the storage pool 41 (volume 42) from the user terminal 10, it changes the capacity of the storage pool 41 in accordance with the change instruction only if the capacity indicated by the change instruction does not exceed the contracted capacity, while in the expansion determination process s51, resources are added without changing the capacity of the storage pool 41, and in the reduction determination process s54, resources are reduced without changing the capacity of the storage pool 41.

[0119] This makes it possible to prevent the storage pool 41 (volume 42) from exceeding the contracted capacity, and also makes it possible to prevent data corruption from occurring when resources are reduced.

[0120] Furthermore, the storage cloud system 40 of this embodiment accepts input of each threshold in the threshold management table 400 .

[0121] This allows appropriate resource changes to be made in accordance with the configuration of the storage pool 41 (volume 42) used by the user.

[0122] The present invention is not limited to the above-described embodiment, and can be implemented using any components without departing from the scope of the present invention. The above-described embodiments and modifications are merely examples, and the present invention is not limited to these contents as long as the features of the invention are not impaired. In addition, although various embodiments and modifications have been described above, the present invention is not limited to these contents. Other aspects conceivable within the scope of the technical idea of ​​the present invention are also included in the scope of the present invention.

[0123] For example, part of the hardware included in each device of this embodiment may be provided in another device.

[0124] Furthermore, each program of each device may be provided in another device, a program may consist of multiple programs, or multiple programs may be integrated into one program.

[0125] Furthermore, the data access performance value may be acquired by a predetermined function provided in advance in the storage cloud system 40 .

[0126] Furthermore, the method of determining the magnitude of the data access load described in this embodiment is merely an example, and various modified examples are possible. [Explanation of symbols]

[0127] 1 Information processing system, 20 Storage system, 40 Storage cloud system, 41 Storage pool, 43 Service portal platform, 45 Storage node group, 46 Storage node

Claims

1. an expansion process for calculating a load value of a storage device having a predetermined cluster configuration, judging whether the calculated load value exceeds a predetermined value, and adding a predetermined resource to the storage device if the calculated load value exceeds the predetermined value; a calculation device that adds the resource and then calculates a predicted value of the load of the storage when the resource is reduced, determines whether the calculated predicted value is below a predetermined value, and if the calculated predicted value is below the predetermined value, executes a reduction process to reduce the resource of the storage.

2. The computing device includes: In the expansion process, a load value of a storage device having a predetermined processing performance and a predetermined cluster configuration is calculated, and it is determined whether the calculated load value exceeds a predetermined value. If the calculated load value exceeds the predetermined value, the processing performance is increased or the number of nodes constituting the cluster of the storage device is increased. in the reduction process, after increasing the processing performance or increasing the number of nodes, calculating the predicted value when the processing performance is decreased or the number of nodes is decreased, determining whether the calculated predicted value is below a predetermined value, and if the calculated predicted value is below the predetermined value, reducing the processing performance of the storage or decreasing the number of nodes of the storage. The storage system according to claim 1 .

3. The computing device includes: In the expansion process, load values ​​relating to data reading and data writing are calculated for a storage device having a predetermined processing performance and a predetermined cluster configuration, and if the calculated load value relating to data reading exceeds a predetermined value, the processing performance is increased, and if the calculated load value relating to data writing exceeds a predetermined value, the number of nodes constituting the cluster of the storage device is increased, In the reduction process, after increasing the processing performance or increasing the number of nodes, a predicted value related to data reading and data writing is calculated in the case where the processing performance is decreased or the number of nodes is decreased, and if the calculated predicted value related to data reading and data writing is below a predetermined value, the processing performance is decreased or the number of nodes is decreased. The storage system according to claim 2.

4. The computing device includes: In the reduction process, when reducing the nodes of the storage device from which the nodes have been increased, the data stored in the storage device from which the nodes have been increased is stored so as to be evenly distributed among the nodes of the storage device from which the nodes have been reduced. The storage system according to claim 3.

5. The computing device includes: In the expansion process, a performance value of data access to the storage device for a predetermined period is calculated, a difference between the calculated performance value and a predetermined threshold is calculated based on an accumulated value for the predetermined period as the load value, and it is determined whether the calculated load value exceeds the predetermined value. The storage system according to claim 1 .

6. The computing device includes: In the expansion process, an accumulated value of the difference between the calculated performance value and the predetermined threshold value during the predetermined period is calculated, a ratio of the calculated accumulated value to an accumulated value of the performance value during the period in which the performance value was the predetermined value is calculated as the load value, and it is determined whether the calculated load value exceeds the predetermined value. The storage system according to claim 5.

7. 3. The storage system according to claim 2, wherein the nodes are configured as information processing devices in a plurality of information processing devices configured to be able to communicate with each other, and the storage is configured as a cluster by the plurality of nodes.

8. The computing device includes: when receiving an instruction to change the capacity of the storage from another information processing device, only if the capacity indicated by the instruction to change does not exceed a predetermined capacity, execute a capacity change process to change the capacity of the storage in accordance with the instruction to change; In the expansion process, the predetermined resource is added without changing the capacity of the storage; In the reduction process, the predetermined resource is reduced without changing the capacity of the storage. The storage system according to claim 1 .

9. The computing device includes: Accepting input of a predetermined value in the expansion process and a predetermined value in the reduction process, The storage system according to claim 1 .

10. An information processing device, an expansion process for calculating a load value for a storage device having a predetermined cluster configuration, judging whether the calculated load value exceeds a predetermined load value, and adding a predetermined resource to the storage device if the calculated load value exceeds the predetermined load value; calculating a load value for the storage device when the resource is added and then reduced, determining whether the calculated load value is below a predetermined load value, and executing a reduction process for reducing the resource of the storage device if the calculated load value is below the predetermined load value. A method for controlling a storage system.

Citation Information

Patent Citations

  • Proportional service quality based on client usage and system metrics.

    JP2015507268A

  • Computer system and method for managing computer resource assignment

    JP2016103113A

  • Information processing system and bursting control method

    JP2022106130A

  • Storage system

    JP2022182584A

  • Computer system, management computer, and management method

    WO2015198441A1