Forecasting storage array capacity

Abstract

An information handling system includes a storage system and a remote processing system. The storage system includes a storage array and a local storage usage predictor. The local storage usage predictor receives usage information from the storage array and predicts a first usage forecast for the storage array based on the usage information. The remote processing system includes a remote storage usage predictor located remote from the storage system. The remote storage usage predictor receives the usage information and predicts a second usage forecast for the storage array based on the usage information.

Description

Technical Field

[0001] The present disclosure relates generally to information handling systems and, more particularly, to storage array capacity in information handling systems. Background Art

[0002] As the value and use of information continues to grow, individuals and businesses seek additional ways to process and store information. One option is an information handling system. An information handling system typically processes, compiles, stores and / or communicates information or data for business, personal or other purposes. Because technology and information handling needs and requirements may vary between different applications, information handling systems may also vary with respect to what information is handled, how the information is handled, how much information is processed, stored or communicated, and how quickly and efficiently the information can be processed, stored or communicated. Variations in information handling systems allow information handling systems to be general or configured for specific users or specific uses (e.g., financial transaction processing, reservations, enterprise data storage, or global communications). In addition, an information handling system may include various hardware and software resources that can be configured to process, store and communicate information, and may include one or more computer systems, data storage systems, and networking systems. Summary of the Invention

[0003] An information handling system may include a storage system and a remote processing system. The storage system includes a storage array and a local storage usage predictor. The local storage usage predictor may receive usage information from the storage array and predict a first usage forecast for the storage array based on the usage information. The remote processing system may include a remote storage usage predictor located remote from the storage system. The remote storage usage predictor may receive the usage information and predict a second usage forecast for the storage array based on the usage information. BRIEF DESCRIPTION OF THE DRAWINGS

[0004] It should be understood that for simplicity and clarity of illustration, the elements shown in the drawings are not necessarily drawn to scale. For example, the dimensions of some elements are exaggerated relative to other elements. Embodiments incorporating the teachings of the present disclosure are shown and described with respect to the drawings presented herein, in which:

[0005] Figure 1 is a block diagram illustrating a storage management system according to an embodiment of the present disclosure;

[0006] Figure 2 is a flow chart illustrating a method for predicting storage capacity in a storage system according to an embodiment of the present disclosure;

[0007] Figure 3 is a flow chart illustrating a method for performing local capacity prediction according to an embodiment of the present disclosure;

[0008] Figure 4 is a flowchart illustrating a method for performing local capacity prediction according to an embodiment of the present disclosure; and

[0009] Figure 5 is a block diagram illustrating a general information handling system according to another embodiment of the present disclosure.

[0010] The use of the same reference symbols in different drawings indicates similar or identical items. DETAILED DESCRIPTION

[0011] The following description, in conjunction with the accompanying drawings, is provided to facilitate understanding of the teachings disclosed herein. The following discussion will focus on specific implementations and embodiments of the teachings. This focus is provided to facilitate description of the teachings and should not be construed as limiting the scope or applicability of the teachings. However, other teachings may certainly be used in this application. These teachings may also be used in other applications and with several different types of architectures, such as distributed computing architectures, client / server architectures, or middleware server architectures and associated resources.

[0012] Figure 1 A storage management system 100 is shown that includes a storage system 110 and a remote storage manager 120. The storage system 110 includes a storage array 112, a local capacity predictor 114, a policy agent 116, a notification manager 118, an event logger 120, and an optimizer 122. The remote storage manager 130 includes a remote capacity predictor 132. In a typical storage management system, an alert is provided when a storage resource, such as a disk drive in a storage array, reaches an alert threshold, such as a storage capacity threshold or a data bandwidth threshold. However, such last-minute alerts typically do not allow sufficient warning time to increase the storage capacity of the storage array to prevent exhaustion of storage space in the storage array. This situation can be alleviated to some extent by setting lower storage capacity limit and data bandwidth limit thresholds. However, setting lower limits may result in an unintended consequence of unnecessarily increasing the number of alerts provided by the storage management system.

[0013] Various machine learning approaches, as well as event and threshold modeling, can be used to improve the predictions provided by alerts. Specifically, simple modeling, such as linear regression, can be used, but such simple modeling is often insufficient to account for diverse scenarios and complex event modeling. More complex modeling can be used to account for a wider range of scenarios and more complex event modeling, but such solutions typically require large amounts of data and processing capacity, limiting the available storage and processing capacity of storage management systems. Furthermore, such complex models typically do not provide real-time predictions.

[0014] In a specific embodiment, a hybrid predictive model is applied to the storage management system 100, wherein a lightweight predictive model is instantiated on the storage system 110 and a more complex, resource-intensive predictive model is instantiated at the remote storage manager 130. Specifically, the local lightweight predictive model instantiated on the storage system 110 utilizes a relatively small historical data set and a simple predictive algorithm to provide real-time, short-term predictions of the data storage capacity of the storage array 112 and the data bandwidth of the storage array. The lightweight predictive model is well suited for providing predictions for emergency use cases so that administrators of the storage management system 100 can take short-term remedial measures, such as powering on additional storage servers, migrating workloads, pausing inactive workloads, and the like.

[0015] The lightweight predictive model can utilize linear regression modeling, autoregression (AR), moving average (MA), autoregressive integrated moving average (ARIMA), etc., which are known to provide satisfactory results for short-term forecasting of demand on the storage management system 100. In a specific embodiment, the lightweight predictive model utilizes a variety of different algorithms to determine the best fit for the data at hand. Thus, events occurring locally on the storage system 100 can be quickly interpreted and incorporated into the determination of the best fit algorithm.

[0016] Examples of local events include the installation or removal of a hard drive, the failure of one or more hard drives, the reallocation of storage partitions, and the like. A forecast schedule can be provided for daily updates, weekly updates, hourly updates, and the like. Therefore, the lightweight forecast model is referred to as "lightweight" both in terms of the type of algorithm used and in terms of the amount of processing resources required by the storage system 110 to implement the algorithm. Consequently, the storage management system 100 achieves good short-term storage forecasting performance without unnecessarily burdening the processing resources of the storage system 110.

[0017] The resource-intensive prediction module is well suited for providing long-term storage forecasts, such as the purchase and installation of additional storage assets. Here, more processing resources external to the storage system 110, such as network or data center servers, cloud processing systems, etc., are utilized to perform more complex predictive processing to more accurately model the behavior of the storage system 110. It should be understood that the processing capacity is larger and the time constraints are looser than those of lightweight prediction models. As needed or desired, the resource-intensive prediction model can utilize long short-term memory (LSTM) modeling, a gradient boosting framework such as XGBoost, or other more resource-intensive prediction models. In addition, the resource-intensive prediction model can utilize a wider range of factors, such as data reduction percentage, number of snapshots, snapshot size, average snapshot retention, number of replication sessions, replication average recovery point objective (RPO), I / O patterns, etc. Here, a prediction timeline can be provided for monthly updates, quarterly updates, etc.

[0018] return Figure 1 Storage array 112 represents the data storage media of storage system 110, such as hard disk drives (HDDs), solid-state drives (SSDs), tape backup, or other storage media as needed or desired. Storage array 112 may represent storage media according to one or more data storage standards, such as SCSI storage devices, Fibre Channel storage devices, etc. Local capacity predictor 114 implements a lightweight forecasting model to generate fast short-term capacity forecasts for storage system 110. Policy agent 116 defines policy conditions for the forecasts provided by local capacity predictor 114.

[0019] Specifically, the policy agent 116 defines the interval for collecting data under the conditions of the storage array 112, the prediction frequency, the default prediction model for performing short-term predictions, a threshold value such as a data capacity threshold or a bandwidth capacity threshold, an optimization strategy such as the number of optimization attempts, the order in which various prediction models are used, the conditions for determining the optimized prediction model, the optimization order, the optimization conditions, etc. Table 1 below provides a table of example policy conditions.

[0020] Table 1 – Example policy conditions

[0021]

[0022] Notification manager 118 is configured to generate notifications based on the short-term forecast. In certain embodiments, notification manager 118 also generates notifications based on the long-term forecast as described above. Event recorder 120 is configured to record events in storage system 110 that affect the local forecast, such as adding or removing storage media from storage array 110, deleting storage media from a storage array for a large storage object, or migrating storage media to a storage array for a large storage object. When the selected model fails the evaluation, optimizer 122 optimizes local capacity predictor 114.

[0023] Specifically, when the selected model fails the evaluation, the optimizer 122 can be used to narrow the scope of the historical data evaluated by the local capacity predictor 114, determine whether a system event has occurred that may have caused the model to fail the evaluation, select a different prediction algorithm, etc., in order to obtain better optimization from the local capacity predictor. The remote capacity predictor 132 implements the resource-intensive prediction model described above.

[0024] Figure 2A method for predicting storage capacity in a storage system is shown, beginning at block 200. In block 202, a short-term storage capacity forecast is provided by a capacity forecaster local to the storage system, while in block 204, a long-term storage capacity forecast is provided by a capacity forecaster remote from the storage system. When, in block 206, the respective forecasts indicate insufficient capacity, notifications based on the forecasts made in blocks 202 and 204 are made, and the method ends in block 208.

[0025] Figure 3 A method for performing local capacity prediction is shown, starting at block 300. In block 302, data for a storage array is collected. For example, storage utilization or available storage utilization, data bandwidth, etc. can be collected based on a policy agent, such as at predetermined intervals for a predetermined duration. In block 204, a local prediction model is established. For example, a default prediction model can include a linear regression algorithm. In block 306, a selected model is evaluated using the collected data. For example, the selected model can be evaluated for a root mean square value (RMS), a root mean square error (RMSE) value, an R2 value, etc. In decision block 308, it is determined whether the model passes the evaluation. For example, the RMS value can be determined to pass the evaluation when the value is within a threshold, where the R2 value is greater than a minimum value, etc.

[0026] If the model passes the evaluation, the "yes" branch of decision block 308 is taken, a short-term capacity forecast is made in block 310, the short-term forecast is passed to the notification manager in block 312, and the method ends in block 314. If the model fails the evaluation, the "no" branch of decision block 308 is taken, and a determination is made in block 316 whether the model matches the policy. For example, the policy agent may determine whether the number of attempts to determine the forecast value exceeds a threshold, or may determine that the processor utilization is less than another threshold. If the model matches the policy, the "yes" branch of decision block 316 is taken, the model is optimized in block 318, and the method returns to block 304. If the model does not match the policy, the "no" branch of decision block 316 is taken, a notification that a forecast was not made is sent in block 320, and the method ends in block 314.

[0027] Figure 4A method for performing local capacity prediction is shown, starting from block 400. In block 402, data for a storage array is collected. In block 204, a remote prediction model is established. For example, a default prediction model may include an LSTM algorithm. In block 406, the selected model is evaluated using the collected data. For example, the selected model may be evaluated for a root mean square value (RMS), a root mean square error (RMSE) value, an R2 value, etc. In decision block 408, it is determined whether the model passes the evaluation. For example, when the value is within a threshold, where the R2 value is greater than a minimum value, etc., it can be determined that the RMS value passes the evaluation. If the model passes the evaluation, the "yes" branch of decision block 408 is taken, a long-term capacity prediction is performed in block 410, the short-term prediction is passed to the notification manager in block 412, and the method ends in block 414.

[0028] If the model fails the evaluation, the "No" branch of decision block 408 is taken, and a determination is made in block 416 as to whether the model matches the policy. For example, the policy agent may determine whether the number of attempts to determine a predicted value exceeds a threshold, or may determine that processor utilization is less than another threshold. If the model does not match the policy, the "No" branch of decision block 416 is taken, the model is optimized in block 418, and the method returns to block 404. If the model does match the policy, the "Yes" branch of decision block 416 is taken, a notification of a missed prediction is sent in block 420, the time interval for the model is updated in block 422, and the method ends in block 414.

[0029] Figure 5 A general embodiment of an information handling system 500 is shown. For the purposes of this disclosure, an information handling system may include any tool or collection of tools that can be used to compute, sort, process, transmit, receive, retrieve, originate, convert, store, display, represent, detect, record, reproduce, dispose of, or utilize any form of information, intelligence, or data for business, scientific, control, entertainment, or other purposes. For example, information handling system 500 may be a personal computer, a laptop computer, a smartphone, a tablet device or other consumer electronic device, a network server, a network storage device, a switching router or another network communication device, or any other suitable device, and may vary in size, shape, performance, functionality, and price.

[0030] In addition, the information handling system 500 may include processing resources for executing machine executable code, such as a central processing unit (CPU), a programmable logic array (PLA), an embedded device such as a system on a chip (SoC), or other control logic hardware. The information handling system 500 may also include one or more computer-readable media for storing machine executable code, such as software or data. Additional components of the information handling system 500 may include one or more storage devices that can store machine executable code, one or more communication ports for communicating with external devices, and various input and output (I / O) devices (such as a keyboard, mouse, and video display). The information handling system 500 may also include one or more buses that can be used to transfer information between the various hardware components.

[0031] The information handling system 500 may include devices or modules embodying one or more devices or modules described below and is configured to perform one or more methods described below. The information handling system 500 includes processors 502 and 504, input / output (I / O) interfaces 510, memories 520 and 525, a graphics interface 530, a basic input and output system / universal extensible firmware interface (BIOS / UEFI) module 540, a disk controller 550, a hard disk drive (HDD) 554, an optical disk drive (ODD) 556, a disk emulator 560 connected to an external solid-state drive (SSD) 562, an I / O bridge 570, one or more additional resources 574, a trusted platform module (TPM) 576, a network interface 580, a management device 590, and a power supply 595. Processors 502 and 504, I / O interface 510, memory 520, graphics interface 530, BIOS / UEFI module 540, disk controller 550, HDD 554, ODD 556, disk emulator 560, SSD 562, I / O bridge 570, additional resources 574, TPM 576, and network interface 580 together provide a host environment for information handling system 500, which is used to provide data processing functions of the information handling system. The host environment is used to execute machine-executable code, including platform BIOS / UEFI code, device firmware, operating system code, application programs, programs, etc., to perform data processing tasks associated with information handling system 500.

[0032] In a host environment, processor 502 is connected to I / O interface 510 via processor interface 506, and processor 504 is connected to the I / O interface via processor interface 508. Memory 520 is connected to processor 502 via memory interface 522. Memory 525 is connected to processor 504 via memory interface 527. Graphics interface 530 is connected to I / O interface 510 via graphics interface 532 and provides video display output 536 to video display 534. In a particular embodiment, information handling system 500 includes separate memories dedicated to each of processors 502 and 504 via separate memory interfaces. Examples of memories 520 and 530 include random access memory (RAM) (e.g., static RAM (SRAM), dynamic RAM (DRAM), non-volatile RAM (NV-RAM), etc.), read-only memory (ROM), another type of memory, or a combination thereof.

[0033] The BIOS / UEFI module 540, the disk controller 550, and the I / O bridge 570 are connected to the I / O interface 510 via an I / O channel 512. Examples of the I / O channel 512 include a peripheral component interconnect (PCI) interface, a PCI expansion (PCI-X) interface, a high-speed PCI-Express (PCIe) interface, another industry standard or proprietary communication interface, or a combination thereof. The I / O interface 510 may also include one or more other I / O interfaces, including an Industry Standard Architecture (ISA) interface, a Small Computer Serial Interface (SCSI) interface, an Inter-Integrated Circuit (IC) interface, and a Serial Bus (SBC) interface. 2 The BIOS / UEFI module 540 includes BIOS / UEFI code that can be used to detect resources within the information handling system 500, provide drivers for resources, initialize resources, and access resources. The BIOS / UEFI module 540 includes code that can be used to detect resources within the information handling system 500, provide drivers for resources, initialize resources, and access resources.

[0034] The disk controller 550 includes a disk interface 552 that connects the disk controller to the HDD 554, the ODD 556, and the disk emulator 560. Examples of the disk interface 552 include an integrated drive electronics (IDE) interface, an advanced technology attachment (ATA) such as a parallel ATA (PATA) interface or a serial ATA (SATA) interface, a SCSI interface, a USB interface, a proprietary interface, or any combination thereof. The disk emulator 560 allows the SSD 564 to be connected to the information handling system 500 via an external interface 562. Examples of the external interface 562 include a USB interface, an IEEE 1394 (FireWire) interface, a proprietary interface, or a combination thereof. Alternatively, the solid state drive 564 can be provided within the information handling system 500.

[0035] I / O bridge 570 includes a peripheral interface 572 that connects the I / O bridge to additional resources 574, TPM 576, and network interface 580. Peripheral interface 572 can be the same type of interface as I / O channel 512, or it can be a different type of interface. Thus, when peripheral interface 572 and the I / O channel are of the same type, I / O bridge 570 expands the capacity of I / O channel 512, and when they are of different types, the I / O bridge converts information from a format suitable for the I / O channel to a format suitable for the peripheral channel 572. Additional resources 574 can include a data storage system, an additional graphics interface, a network interface card (NIC), a sound / video processing card, another additional resource, or a combination thereof. Additional resources 574 can be located on the main circuit board, on a separate circuit board or add-in card disposed within information handling system 500, on a device external to the information handling system, or a combination thereof.

[0036] The network interface 580 represents a NIC that is disposed within the information handling system 500, is located on the main circuit board of the information handling system, is integrated into another component such as the I / O interface 510, is in another suitable location, or a combination thereof. The network interface device 580 includes network channels 582 and 584 that provide interfaces for devices external to the information handling system 500. In a specific embodiment, the network channels 582 and 584 are of a different type from the peripheral channels 572, and the network interface 580 converts information from a format suitable for the peripheral channels to a format suitable for the external device. Examples of network channels 582 and 584 include InfiniBand channels, Fibre Channel channels, Gigabit Ethernet channels, proprietary channel architectures, or a combination thereof. Network channels 582 and 584 can be connected to external network resources (not shown). Network resources can include another information handling system, a data storage system, another network, a grid management system, another suitable resource, or a combination thereof.

[0037] Management device 590 represents one or more processing devices, such as a dedicated baseboard management controller (BMC), a system-on-chip (SoC) device, one or more associated memory devices, one or more network interface devices, a complex programmable logic device (CPLD), and the like, which together provide a management environment for information handling system 500. Specifically, management device 590 connects to various components of the host environment via various internal communication interfaces (e.g., a low pin count (LPC) interface, an inter-integrated circuit (I2C) interface, a PCIe interface, and the like) to provide an out-of-band (OOB) mechanism for retrieving information related to the operation of the host environment, providing BIOS / UEFI or system firmware updates, and managing non-processing components of information handling system 500, such as system cooling fans and power supplies. Management device 590 may include a network connection to an external management system, and the management device may communicate with the management system to report status information of information handling system 500, receive BIOS / UEFI or system firmware updates, or perform other tasks for managing and controlling the operation of information handling system 500. The management device 590 can operate outside of a power plane separate from the components of the host environment, such that when the information handling system is otherwise shut down, the management device receives power to manage the information handling system 500. Examples of the management device 590 include commercially available BMC products or other devices that operate according to the Intelligent Platform Management Initiative (IPMI) specification, the Web Services Management (WSMan) interface, the Redfish application programming interface (API), another Distributed Management Task Force (DMTF), or other management standards, and may include an integrated Dell Remote Access Controller (iDRAC), an embedded controller (EC), etc. The management device 590 may also include associated memory devices, logic devices, security devices, etc., as needed or desired.

[0038] Although only a few exemplary embodiments have been described in detail herein, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without materially departing from the novel teachings and advantages of the embodiments of the present disclosure. Therefore, all such modifications are intended to be included within the scope of the embodiments of the present disclosure as defined by the appended claims. In the claims, means-plus-function clauses are intended to cover the structures described herein as performing the recited function and not only structural equivalents but also equivalent structures.

[0039] The subject matter disclosed above is to be considered illustrative and not restrictive, and the appended claims are intended to cover any and all such modifications, enhancements, and other embodiments that fall within the scope of the invention. Thus, to the maximum extent allowed by law, the scope of the present invention is to be determined by the broadest permissible interpretation of the appended claims and their equivalents, and shall not be restricted or limited by the foregoing detailed description.

Claims

1. An information processing system comprising: a storage system comprising a storage array and a local storage usage predictor configured to receive usage information from the storage array and predict a first usage prediction for the storage array based on the usage information; as well as A remote processing system includes a remote storage usage predictor located remote from the storage system, the remote storage usage predictor configured to receive the usage information and predict a second usage prediction for the storage array based on the usage information.

2. The information handling system of claim 1, wherein the local storage usage predictor predicts the first usage prediction based on at least one of a linear regression model, an autoregressive model, a moving average model, and an autoregressive integrated moving average model.

3. The information handling system of claim 2, wherein the local storage usage predictor is further configured to utilize each of the linear regression model, the autoregressive model, the moving average model, and the autoregressive integrated moving average model in predicting the first usage prediction.

4. The information handling system of claim 2 , wherein the local storage usage predictor is further configured to utilize the linear regression model as a default prediction model when predicting the first usage prediction; determine that the first usage prediction fails an evaluation criterion; and utilize a second model of the autoregressive model, the moving average model, and the autoregressive integrated moving average model to predict a third usage prediction for the storage array based on the usage information.

5. The information handling system of claim 1, wherein the local storage usage predictor is further configured to determine an event associated with the storage array, wherein the first usage prediction is further based on the event.

6. The information handling system of claim 5, wherein the event comprises one of: an increase in storage capacity of the storage array, a decrease in the storage capacity of the storage array, migration of an image object into the storage array, and migration of the image object out of the storage array. 7 . The information handling system of claim 1 , wherein the remote storage usage predictor predicts the long-term usage prediction based on at least one of a long short-term memory model and a gradient boosting framework model.

8. The information handling system of claim 1, wherein the usage information comprises one of a current storage capacity of the storage array and a current bandwidth of the storage array.

9. The information handling system of claim 8, wherein the first usage prediction comprises one of a storage capacity prediction and a data bandwidth prediction of the storage array.

10. The information handling system of claim 1, wherein the first usage scenario is predicted to have a shorter duration than the second usage scenario is predicted to have a shorter duration.

11. A method comprising: receiving, by a local storage usage predictor of the storage system, usage information from a storage array of the storage system; predicting, by the local storage usage predictor, a first usage forecast for the storage array based on the usage information; receiving the usage information via a remote storage usage predictor remote from the storage system; as well as A second usage forecast for the storage array is predicted by the remote storage usage forecaster based on the usage information, wherein a duration of the first usage forecast is shorter than a duration of the second usage forecast.

12. The method of claim 11, wherein the local storage usage predictor predicts the first usage prediction based on at least one of a linear regression model, an autoregressive model, a moving average model, and an autoregressive integrated moving average model.

13. The method of claim 12, further comprising: Each of the linear regression model, the autoregressive model, the moving average model, and the autoregressive integrated moving average model is utilized by the local storage usage predictor in predicting the first usage prediction.

14. The method of claim 12, further comprising: utilizing, by the local storage usage predictor, the linear regression model as a default prediction model when predicting the first usage prediction; determining that the first usage prediction fails an evaluation criterion; and A second model of the autoregressive model, the moving average model, and the autoregressive integrated moving average model is utilized to predict a third usage prediction of the storage array based on the usage information.

15. The method of claim 11, further comprising: An event associated with the storage array is determined by the local storage usage predictor, wherein the first usage prediction is further based on the event.

16. The method of claim 15, wherein the event comprises one of: an increase in storage capacity of the storage array, a decrease in the storage capacity of the storage array, migration of an image object into the storage array, and migration of the image object out of the storage array.

17. The method of claim 11, wherein the remote storage usage predictor predicts the long-term usage prediction based on at least one of a long short-term memory (LSTM) model and a gradient boosting framework model.

18. The method of claim 11, wherein the usage information comprises one of a current storage capacity of the storage array and a current bandwidth of the storage array.

19. The method of claim 18, wherein the first usage forecast comprises one of a storage capacity forecast and a data bandwidth forecast of the storage array.

20. An information processing system comprising: Notification Manager; a storage system comprising a storage array and a local storage usage predictor configured to receive usage information from the storage array, predict a first usage prediction for the storage array based on the usage information, and send the first usage prediction to the notification manager; as well as a remote processing system comprising a remote storage usage predictor remote from the storage system, the remote storage usage predictor configured to receive the usage information, predict a second usage prediction for the storage array based on the usage information, and send the second usage prediction to the notification manager; Wherein the notification manager is configured to provide a first notification based on the first usage prediction and to provide a second notification based on the second usage prediction.

Images