Solid state disk closed loop self-healing test method and test system

Abstract

The application discloses a solid state disk closed loop self-healing test method and a test system. The method comprises the following steps: injecting data loads with different compression characteristics into a solid state disk to be tested, collecting performance parameters and medium operation parameters corresponding to each data load; continuously applying a first load to the solid state disk and superimposing a second load to form a composite aging load; pausing the composite aging load, re-injecting the data loads with different compression characteristics into the solid state disk to retest the current write amplification factor, and determining the abnormality level corresponding to the current abnormality; when the abnormality level is a correctable abnormality level, performing a corresponding self-healing processing operation and continuing the test after the self-healing processing is completed; when the abnormality level is an uncorrectable abnormality level, terminating the test and recording the logical address interval with errors. The application can simulate the load characteristics of burst superposition, perform corresponding self-healing processing operations according to the abnormality level, and realize automatic testing while improving the test efficiency.

Description

Technical Field

[0001] This application belongs to the field of solid-state drive technology, specifically relating to a closed-loop self-healing test method and test system for solid-state drives. Background Technology

[0002] With the rapid development of cloud computing, big data, and artificial intelligence, solid-state drives (SSDs) have been widely used in various computing devices due to their advantages such as high-speed read / write speeds and low power consumption. Their long-term operational reliability directly affects the stability and reliability of storage. During product development, verification, and mass production testing, aging tests are a core method for assessing the health of SSDs. The degree to which the test scenario closely matches actual operating conditions and the level of automation in the testing process directly determine the accuracy of product quality assessment and overall testing efficiency.

[0003] Currently, conventional aging tests mostly use a fixed load mode, which continuously reads and writes to the solid-state drive under a preset single load throughout the entire test cycle. This makes it difficult to reproduce the load characteristics superimposed in a real environment, resulting in a deviation between the test results and the actual operating performance.

[0004] Furthermore, existing aging tests are mostly open-loop processes, which are prone to interruption when performance fluctuations or media abnormalities occur during the test. They rely on manual intervention for judgment and recovery, resulting in low test efficiency and making it difficult to meet the requirements for test continuity in batch testing scenarios. Summary of the Invention

[0005] This application provides a closed-loop self-healing test method and system for solid-state drives, which can simulate sudden superposition of load characteristics and perform corresponding self-healing operations according to the anomaly level, thereby achieving automated testing while improving testing efficiency.

[0006] To address the aforementioned technical problems, this application provides a closed-loop self-healing testing method for solid-state drives, comprising the following steps: Data payloads with different compression characteristics are injected into the solid-state drive under test, and the performance parameters and media operation parameters corresponding to each data payload are collected to construct an initial benchmark dataset that includes at least the write amplification factor baseline. A first load is continuously applied to the solid-state drive, and without interrupting the first load, a second load with a random access intensity higher than the first load is periodically superimposed to form a composite aging load. After completing the second load superposition for a preset number of times, the composite aging load is paused. Under the same test conditions as when constructing the write amplification factor baseline, the data load with different compression characteristics is re-injected into the solid-state drive to retest the current write amplification factor. Real-time collection of solid-state drive performance data and media health data, combined with the initial benchmark dataset for anomaly detection, to determine the anomaly level corresponding to the current anomaly; When the anomaly level is a correctable anomaly level, the self-healing operation corresponding to the anomaly level is executed, and the test continues after the self-healing operation is completed. When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

[0007] As a further improvement of this application, the data payload with different compression characteristics includes a first data payload and a second data payload, wherein the first data payload is an incompressible data stream and the second data payload is a compressible data stream; The write amplification factor baseline in the initial benchmark dataset includes a first write amplification factor baseline corresponding to the first data load and a second write amplification factor baseline corresponding to the second data load, wherein the first write amplification factor baseline is different from the second write amplification factor baseline.

[0008] As a further improvement to this application, determining the anomaly level corresponding to the current anomaly includes: When the performance metrics in the performance data deviate from the performance baseline values ​​in the initial benchmark dataset, and the media error count in the media health data does not increase, it is determined to be at the first abnormal level. When the media error count in the media health data increases and no correctable input / output error event is captured during the execution of test read / write operations, it is determined to be at the second abnormal level; When a correctable input / output error event is captured during the execution of a test read / write operation, it is classified as a third-level anomaly. When an uncorrectable error status code is received from the solid-state drive, it is determined to be at the fourth level of abnormality. Among them, the first abnormality level, the second abnormality level, and the third abnormality level are correctable abnormality levels, while the fourth abnormality level is an uncorrectable abnormality level.

[0009] As a further improvement to this application, when the anomaly level is the first anomaly level, the execution of the self-healing operation corresponding to the anomaly level includes: Reduce the queue depth of the first load and extend the total test duration to keep the total amount of data written to the composite aging load constant.

[0010] As a further improvement to this application, when the anomaly level is the second anomaly level, the execution of the self-healing processing operation corresponding to the anomaly level includes: Reduce the queue depth of the first load; A full disk read scan is performed on the solid-state drive to trigger the background inspection and bad block management mechanism of the solid-state drive firmware; After the scan is complete, restore the queue depth of the first load.

[0011] As a further improvement to this application, when the anomaly level is the third anomaly level, the execution of the self-healing processing operation corresponding to the anomaly level includes: Save the current test progress and load configuration status; A space reclamation command is issued to the logical address range associated with the correctable input / output error event, causing the firmware of the solid-state drive to reclaim the physical resources mapped by the logical address range and trigger bad block remapping processing. After the space reclamation command is executed, the test is resumed based on the saved test progress and load configuration status.

[0012] As a further improvement to this application, after determining the anomaly level corresponding to the current anomaly, the method further includes: When the performance metrics in the performance data decrease: If the fluctuation range of the first write amplification factor obtained by retesting relative to the baseline of the first write amplification factor and the fluctuation range of the second write amplification factor obtained by retesting relative to the baseline of the second write amplification factor are both within the preset range, the temperature of the solid-state drive exceeds the preset temperature threshold, and the media error count in the media health data does not increase, then the cause of the performance degradation of the solid-state drive is determined to be thermal throttling. If the fluctuation range of the first write amplification factor obtained by retesting relative to the baseline of the first write amplification factor and the fluctuation range of the second write amplification factor obtained by retesting relative to the baseline of the second write amplification factor are both within the preset range, the temperature of the solid-state drive does not exceed the preset temperature threshold, and the available spare blocks in the media health data decrease or the media error count increases, then it is determined that the cause of the performance degradation of the solid-state drive is the increase of bad blocks in the media. If the difference between the first write amplification factor obtained from the retest and the second write amplification factor obtained from the retest exceeds a preset threshold relative to the difference between the baseline of the first write amplification factor and the baseline of the second write amplification factor, and the temperature of the solid-state drive does not exceed a preset temperature threshold, and the media error count in the media health data does not increase, then the cause of the solid-state drive performance degradation is determined to be the degradation of the main control compression engine.

[0013] As a further improvement to this application, after real-time acquisition of the performance data and media health data of the solid-state drive, the method further includes: Based on the real-time collected performance data and media health data of the solid-state drive, the probability of failure within a future preset time window is obtained through time-series analysis. When the failure probability exceeds a preset probability threshold, the queue depth of the first load is proactively reduced before determining the anomaly level corresponding to the current anomaly.

[0014] As a further improvement to this application, the queue depth of the second load is higher than the queue depth of the first load; and / or, The read / write addresses of the first load are distributed sequentially, while the read / write addresses of the second load are distributed randomly.

[0015] As a further improvement to this application, this application also provides a closed-loop self-healing testing system for solid-state drives, comprising: The acquisition unit is used to inject data payloads with different compression characteristics into the solid-state drive under test, collect the performance parameters and media operation parameters corresponding to each data payload, and construct an initial benchmark dataset that includes at least the write amplification factor baseline. The superposition unit is used to continuously apply a first load to the solid-state drive, and periodically superimpose a second load with a random access intensity higher than the first load without interrupting the first load, so as to form a composite aging load. The retesting unit is used to pause the composite aging load after completing a preset number of second load superpositions, and re-inject the data load with different compression characteristics into the solid-state drive under the same test conditions as when constructing the write amplification factor baseline to retest the current write amplification factor. The detection unit is used to collect the performance data and media health data of the solid-state drive in real time, and combine them with the initial benchmark dataset to perform anomaly detection and determine the anomaly level corresponding to the current anomaly. The processing unit is configured to, when the anomaly level is a correctable anomaly level, execute a self-healing operation corresponding to the anomaly level, and continue testing after the self-healing operation is completed; and, When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

[0016] This application provides a closed-loop self-healing testing method and system for solid-state drives (SSDs). Before aging testing, data loads with different compression characteristics are injected into the SSD under test. Performance parameters and media operating parameters corresponding to each data load are collected to construct an initial benchmark dataset that includes at least a write amplification factor baseline. During the aging test phase, a first load is continuously applied to the SSD under test, and a second load with higher random access intensity is periodically superimposed without interrupting the first load, forming a composite aging load. The SSD under test is subjected to continuous steady-state read / write stress while being periodically impacted by a high random access intensity load, in order to more closely approximate the real-world usage environment of sudden traffic surges.

[0017] After completing the second load stacking for a preset number of cycles, the composite aging load is paused. Under the same test conditions as when building the write amplification factor baseline, a data load with different compression characteristics is re-injected to retest the current write amplification factor. Real-time performance data and media health data of the tested SSD are collected and combined with the initial benchmark dataset for anomaly detection and anomaly level determination. When the anomaly level is correctable, the corresponding self-healing operation is executed, and testing continues after completion, preventing direct interruption due to minor anomalies and achieving automatic continuation of the test process. When the anomaly level is uncorrectable, the test is terminated, and the logical address range of the error is recorded, providing location information for subsequent fault analysis. This application can simulate the characteristics of sudden load stacking and execute corresponding self-healing operations according to the anomaly level, achieving automated testing while improving testing efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only a part of the embodiments of this application, and not all of the embodiments. For those skilled in the art, other drawings obtained from these drawings without creative effort are all within the scope of protection of this application.

[0019] Figure 1 A flowchart of a closed-loop self-healing test method for solid-state drives provided in an embodiment of this application.

[0020] Figure 2 This is a flowchart illustrating the determination of fault probability in the solid-state drive closed-loop self-healing test method provided in this application embodiment.

[0021] Figure 3 This is a flowchart illustrating the determination of anomaly levels in the closed-loop self-healing test method for solid-state drives provided in this application embodiment.

[0022] Figure 4 This is a flowchart illustrating the process of determining the cause of solid-state drive performance degradation in the closed-loop self-healing test method for solid-state drives provided in this application embodiment.

[0023] Figure 5 This is a flowchart of the self-healing process corresponding to the first abnormality level in the solid-state drive closed-loop self-healing test method provided in this application embodiment.

[0024] Figure 6 This is a flowchart of the self-healing process corresponding to the second abnormality level in the solid-state drive closed-loop self-healing test method provided in this application embodiment.

[0025] Figure 7 This is a flowchart of the self-healing process corresponding to the third abnormality level in the solid-state drive closed-loop self-healing test method provided in this application embodiment.

[0026] Figure 8 This is a functional block diagram of the solid-state drive closed-loop self-healing test system provided in an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0028] To make the description of this disclosure more detailed and complete, illustrative descriptions of the implementation methods and specific embodiments of this application are provided below; however, this is not the only form of implementing or utilizing the specific embodiments of this application. The implementation methods cover the features of multiple specific embodiments and the method steps and their order for constructing and operating these specific embodiments. However, other specific embodiments can also be used to achieve the same or equivalent functions and step sequences. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in sequences other than those illustrated or described herein.

[0030] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The word "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more. Other quantifiers should be understood similarly. The preferred embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application. Furthermore, the embodiments of this application and the features in the embodiments can be combined with each other without conflict.

[0031] With the rapid development of cloud computing, big data, and artificial intelligence, solid-state drives (SSDs) have been widely used in various computing devices due to their advantages such as high-speed read / write speeds and low power consumption. Their long-term operational reliability directly affects the stability and reliability of storage. During product development, verification, and mass production testing, aging tests are a core method for assessing the health of SSDs. The degree to which the test scenario closely matches actual operating conditions and the level of automation in the testing process directly determine the accuracy of product quality assessment and overall testing efficiency.

[0032] Currently, conventional aging tests mostly use a fixed load mode, which continuously reads and writes to the solid-state drive under a preset single load throughout the entire test cycle. This makes it difficult to reproduce the load characteristics superimposed in a real environment, resulting in a deviation between the test results and the actual operating performance.

[0033] Furthermore, existing aging tests are mostly open-loop processes, which are prone to interruption when performance fluctuations or media abnormalities occur during the test. They rely on manual intervention for judgment and recovery, resulting in low test efficiency and making it difficult to meet the requirements for test continuity in batch testing scenarios.

[0034] In view of this, please refer to Figures 1-8 This application proposes a closed-loop self-healing test method and system for solid-state drives, which can simulate the characteristics of sudden superposition of loads and perform corresponding self-healing operations according to the anomaly level, thereby achieving automated testing while improving testing efficiency.

[0035] Please refer to Figure 1 This is a flowchart of a closed-loop self-healing test method for solid-state drives provided in this application embodiment. The closed-loop self-healing test method for solid-state drives includes the following steps: Step S1: Inject data payloads with different compression characteristics into the solid-state drive under test, collect the performance parameters and media operation parameters corresponding to each data payload, and construct an initial benchmark dataset that includes at least the write amplification factor baseline. In this embodiment, the test host and the load generation engine installed on it constitute the test environment, and the solid-state drive under test is connected to the test host through a standard storage protocol interface. The solid-state drive under test has a main controller and a NAND flash memory array inside. The main controller is responsible for managing data transmission, data compression, and bad block management between the test host and the NAND flash memory.

[0036] Furthermore, the test host injects data payloads with different compression characteristics into the SSD under test via a load generation engine. One type of data payload is almost uncompressed during processing by the SSD's main controller, while the other type is significantly compressed. Since the actual amount of data written to the SSD differs depending on the compression characteristics of the data payloads, injecting these two types of data payloads allows us to observe the SSD's operating state under two different boundary conditions.

[0037] Furthermore, during the injection of each data payload, the test host synchronously collects the corresponding performance parameters and media operating parameters. The performance parameters are the read / write response metrics of the tested SSD on the test host side, while the media operating parameters are the internal operating parameters obtained from the tested SSD through standard storage protocols, specifically including the actual amount of data written to the tested SSD's internal NAND (NAND Flash) flash memory.

[0038] In an optional embodiment, it is necessary to calculate the ratio of the actual amount of data written to the NAND flash memory inside the solid-state drive under test to the amount of data written to the solid-state drive under test by the test host, and use the calculated ratio as the write amplification factor baseline, thereby constructing an initial benchmark dataset that at least includes the write amplification factor baseline.

[0039] As an optional implementation, the data payload with different compression characteristics includes a first data payload and a second data payload, wherein the first data payload is an incompressible data stream and the second data payload is a compressible data stream; In this embodiment, data payloads with different compression characteristics are specifically divided into a first data payload and a second data payload. The first data payload is an incompressible data stream, consisting of a random binary sequence generated by a random number generator, which the host controller of the tested solid-state drive cannot effectively compress. The second data payload is a compressible data stream, consisting of a data sequence of specific repeating characters, which the host controller of the tested solid-state drive can compress at a higher ratio.

[0040] Specifically, the test host injects a first data payload and a second data payload into the solid-state drive under test, and collects the performance parameters and media operating parameters corresponding to the two data payloads, respectively. The first data payload corresponds to the boundary condition where the main controller has no compression effect, and the second data payload corresponds to the boundary condition where the main controller has compression effect.

[0041] As an optional implementation, the write amplification factor baseline in the initial benchmark dataset includes a first write amplification factor baseline corresponding to the first data load and a second write amplification factor baseline corresponding to the second data load, wherein the first write amplification factor baseline is different from the second write amplification factor baseline.

[0042] It is understandable that, since the first data payload is an incompressible data stream, the main controller of the tested solid-state drive cannot compress it. When the first data payload is injected, the amount of data actually written by the main controller to the NAND flash memory is basically the same as the amount of data sent by the test host. The baseline of the first write amplification factor calculated accordingly reflects the write amplification level of the main controller under no compression effect.

[0043] The second data payload is a compressible data stream. After the main controller of the solid-state drive under test compresses it, the actual amount of data written to the NAND flash memory is less than the amount of data sent by the test host. Consequently, the calculated second write amplification factor baseline will be less than the first write amplification factor baseline.

[0044] Furthermore, the first write amplification factor baseline and the second write amplification factor baseline are both constructed into the initial benchmark dataset to provide a comparative basis for evaluating the changes in the data processing capability of the main controller of the tested solid-state drive in subsequent aging tests.

[0045] Step S2: Continuously apply a first load to the solid-state drive, and periodically add a second load with a random access intensity higher than the first load without interrupting the first load, so as to form a composite aging load; In this embodiment, the test host continuously applies a first load to the solid-state drive under test through a load generation engine. The first load is a basic read / write load that operates with a stable read / write ratio and address access mode, used to simulate the continuous read / write stress that the solid-state drive under test would experience under normal business scenarios, thus providing continuous aging conditions for the solid-state drive under test.

[0046] During the application of the first load, the test host does not interrupt the operation of the first load, and adds a second load on top of the first load according to a preset time period. The random access intensity of the second load is higher than that of the first load.

[0047] As can be understood, random access strength is an indicator of the randomness of read and write operations on NAND flash memory. The higher the random access strength, the more dispersed and irregular the read and write operations are on the various physical locations of the NAND flash memory. Since the second load has a higher random access strength, its read and write operations will cover a wide range of the tested solid-state drive, causing storage cells that were not fully accessed under the uniform first load to also be subjected to read and write stress.

[0048] Furthermore, the first load and the periodically superimposed second load together form a composite aging load. Under the composite aging load, the tested solid-state drive simultaneously bears the continuous steady-state stress from the first load and the periodic high-intensity random stress from the second load, which is closer to the actual working conditions of background read and write traffic and sudden high load superposition in real business environments.

[0049] In an optional embodiment, the queue depth of the second load is higher than the queue depth of the first load; and / or, The read / write addresses of the first load are distributed sequentially, while the read / write addresses of the second load are distributed randomly.

[0050] In this embodiment, the random access strength of the second load is higher than that of the first load. Specifically, the queue depth of the second load can be set to be higher than that of the first load, or the read and write addresses of the first load can be set to a sequential address distribution and the read and write addresses of the second load can be set to a random address distribution.

[0051] Queue depth refers to the number of read / write commands waiting to be processed or being processed on the tested SSD at the same time. The second load uses a higher queue depth than the first load, meaning that more read / write commands are sent to the tested SSD at the same time. The tested SSD's main controller needs to process more concurrent requests in a shorter time, thus generating higher processing pressure.

[0052] The read / write address distribution pattern refers to the distribution pattern of logical addresses accessed by read / write commands within the address space. The first workload exhibits a sequential address distribution, where read / write commands access adjacent logical address spaces in ascending or descending order of address, resulting in relatively concentrated access to NAND flash memory. The second workload exhibits a random address distribution, where read / write commands access logical addresses randomly across the entire address space of the tested SSD, leading to more dispersed access to different physical locations within the NAND flash memory and thus covering storage cell areas that are difficult to access under a sequential address distribution.

[0053] Step S3: After completing the second load superposition for a preset number of times, pause the composite aging load, and under the same test conditions as when constructing the write amplification factor baseline, re-inject the data load with different compression characteristics into the solid-state drive to retest the current write amplification factor; In this embodiment of the application, after the solid-state drive under test has been running under a composite aging load for a period of time, its internal NAND flash memory array has been subjected to a second load excitation of continuous read and write and periodic high random access intensity, and the wear state of the storage cells and the data processing capability of the main controller may change.

[0054] When the number of times the second load is stacked reaches the preset number, the test host pauses the composite aging load. During the pause, both the first load and the second load stop sending read and write commands to the solid-state drive under test. At this time, the solid-state drive under test is in an idle state without external read and write pressure.

[0055] After pausing the composite aging load, the test host re-injects the data payload with different compression characteristics used in step S1 into the SSD under the same test conditions as when building the write amplification factor baseline. Here, "same test conditions" means that the load configuration parameters and ambient temperature used when injecting the data payload are consistent with those used when building the write amplification factor baseline.

[0056] Furthermore, the test host collects the performance parameters and media operating parameters of the tested SSD during the re-injection process, and calculates the current write amplification factor, i.e., the current write amplification factor obtained from the retest. This current write amplification factor reflects the ability of the tested SSD's main controller to process data loads with different compression characteristics after undergoing a period of compound aging.

[0057] Step S4: Collect the performance data and media health data of the solid-state drive in real time, combine them with the initial benchmark dataset to perform anomaly detection, and determine the anomaly level corresponding to the current anomaly; In this embodiment, the test host collects performance data and media health data of the solid-state drive under test in real time during compound aging load operation. Performance data refers to observable read / write response metrics on the test host side, while media health data consists of parameters related to the storage media status obtained from the solid-state drive under test using standard storage protocols.

[0058] Furthermore, the test host will compare and analyze the real-time performance data and media health data collected with the initial benchmark dataset constructed in step S1. The initial benchmark dataset contains the performance baseline value and media operating parameter baseline value of the tested SSD before the aging test begins. By comparing the currently collected real-time data with the initial benchmark dataset, the anomaly level corresponding to the current anomaly can be further determined. Different anomaly levels correspond to different degrees of severity of the tested SSD's operating state.

[0059] As an optional implementation method, please refer to Figure 2 This is a flowchart illustrating the process of determining the failure probability in the solid-state drive closed-loop self-healing test method provided in this application embodiment. After real-time acquisition of the solid-state drive's performance data and media health data, the method further includes: Step S40: Based on the real-time collected performance data and media health data of the solid-state drive, the probability of failure within a future preset time window is obtained through time series analysis; Step S41: When the failure probability exceeds a preset probability threshold, before determining the anomaly level corresponding to the current anomaly, actively reduce the queue depth of the first load.

[0060] In this embodiment, after real-time acquisition of performance data and media health data of the tested solid-state drive (SSD), the test host performs time-series analysis on the acquired real-time performance data and real-time media health data. Time-series analysis involves extracting trends and recognizing patterns from data sequences arranged in chronological order to determine the evolution trends of the tested SSD's performance indicators and media health status over time, thereby predicting the probability of the tested SSD failing within a preset time window.

[0061] It should be noted that extracting trends and identifying patterns from performance data and media health data arranged in chronological order, and then predicting future states based on the evolution patterns of historical data, are existing techniques in the field of data analysis. For example, the above time series analysis process can be achieved based on curve fitting methods of statistical regression or trend extrapolation methods based on sliding windows. Therefore, this application will not elaborate on this further.

[0062] Furthermore, if the obtained failure probability exceeds a preset probability threshold, it indicates that the tested SSD is highly likely to experience an anomaly within a preset time window. In this case, before officially determining the anomaly level corresponding to the current anomaly, the test host can proactively reduce the queue depth of the first load, thereby reducing the processing pressure on the tested SSD and taking intervention measures in advance before the tested SSD actually fails.

[0063] As an optional implementation method, please refer to Figure 3This is a flowchart illustrating the process of determining the anomaly level in the closed-loop self-healing test method for solid-state drives provided in this application embodiment. Determining the anomaly level corresponding to the current anomaly includes: Step S42: When the performance indicators in the performance data deviate from the performance baseline values ​​in the initial benchmark dataset, and the media error count in the media health data does not increase, it is determined to be at the first abnormal level; Step S43: When the media error count in the media health data increases and no correctable input / output error event is captured during the test read / write operation, it is determined to be at the second abnormal level; Step S44: When a correctable input / output error event is captured during the execution of a test read / write operation, it is determined to be of the third abnormal level; Step S45: When an uncorrectable error status code is received from the solid-state drive, it is determined to be the fourth abnormal level; Among them, the first abnormality level, the second abnormality level, and the third abnormality level are correctable abnormality levels, while the fourth abnormality level is an uncorrectable abnormality level.

[0064] In this embodiment, the test host determines the level of the current anomaly based on the collected performance data and media health data. The performance baseline value is a component of the initial benchmark dataset constructed in step S1, reflecting the normal performance level of the tested solid-state drive before the aging test begins. The media error count is the cumulative number of storage media errors reported by the tested solid-state drive through the SMART (Self-Monitoring, Analysis and Reporting Technology) attribute, and can be obtained directly.

[0065] In an optional embodiment, when the performance metric in the performance data deviates from the performance baseline value and the media error count does not increase, it indicates that the read and write response capability of the tested solid-state drive has decreased, but the NAND flash memory storage medium itself has not generated new errors. This is judged as the first abnormal level, which corresponds to the situation where there is only a deviation at the performance level while the storage medium is still normal.

[0066] In another optional embodiment, when the media error count increases, but the test host does not capture any correctable input / output error events during the test read / write operation, it indicates that some storage cells in the NAND flash memory have errors, but these errors have not yet had a captureable impact on the test host's read / write operation. This is judged as the second abnormal level, which corresponds to the situation where the storage medium has errors but the errors have not yet affected the upper layer read / write operation.

[0067] In another optional embodiment, when the test host captures a correctable input / output error event during the execution of test read / write operations, it indicates that the error in the storage medium has affected the read / write operations of the test host, and the correctable input / output error event carries the logical address information of the error. At this time, it is determined to be the third abnormal level, which corresponds to the situation where the error can be perceived by the test host and the specific logical address can be located.

[0068] In another optional embodiment, when the test host receives an uncorrectable error status code reported by the solid-state drive under test, it indicates that the error correction mechanism inside the solid-state drive under test can no longer recover the data error. At this time, it is judged as the fourth abnormal level. Here, the uncorrectable error status code is the error information returned by the solid-state drive under test to the test host through the standard storage protocol.

[0069] Specifically, the first, second, and third anomaly levels mentioned above are all correctable anomaly levels. Under these anomaly levels, the tested solid-state drive may still be able to recover to normal operation by adjusting the test load or triggering internal repair operations. The fourth anomaly level is an uncorrectable anomaly level. Under this level, the tested solid-state drive has suffered data corruption that cannot be automatically recovered.

[0070] As an optional implementation method, please refer to Figure 4 This is a flowchart illustrating the process of determining the cause of solid-state drive performance degradation in the solid-state drive closed-loop self-healing test method provided in this application embodiment. After determining the anomaly level corresponding to the current anomaly, the method further includes: When the performance metrics in the performance data decrease: Step S460: If the fluctuation range of the first write amplification factor obtained by retesting relative to the first write amplification factor baseline and the fluctuation range of the second write amplification factor obtained by retesting relative to the second write amplification factor baseline are both within the preset range, the temperature of the solid-state drive exceeds the preset temperature threshold, and the media error count in the media health data does not increase, then it is determined that the cause of the performance degradation of the solid-state drive is thermal throttling. Step S461: If the fluctuation range of the first write amplification factor obtained by retesting relative to the first write amplification factor baseline and the fluctuation range of the second write amplification factor obtained by retesting relative to the second write amplification factor baseline are both within the preset range, the temperature of the solid-state drive does not exceed the preset temperature threshold, and the available spare blocks in the media health data decrease or the media error count increases, then it is determined that the cause of the performance degradation of the solid-state drive is the increase of bad blocks in the media. Step S462: If the difference between the first write amplification factor obtained from the retest and the second write amplification factor obtained from the retest exceeds a preset threshold relative to the difference between the baseline of the first write amplification factor and the baseline of the second write amplification factor, the temperature of the solid-state drive does not exceed the preset temperature threshold, and the media error count in the media health data does not increase, then it is determined that the cause of the performance degradation of the solid-state drive is the degradation of the main control compression engine.

[0071] In this embodiment of the application, after determining the anomaly level corresponding to the current anomaly in step S4, when the performance indicators in the performance data show a decline, the test host will combine data from multiple dimensions to further determine the specific reasons for the performance degradation of the solid-state drive.

[0072] The first write amplification factor baseline is a component of the initial benchmark dataset constructed in step S1, corresponding to the write amplification level of the first data load under the condition that the main controller of the SSD under test has no compression effect. The second write amplification factor baseline is also constructed in step S1, corresponding to the write amplification level of the second data load under the condition that the main controller of the SSD under test has compression effect. The number of available spare blocks is the number of normal storage cells reserved in the NAND flash array inside the SSD under test for replacing bad blocks, which belongs to the media health parameter in the SMART attribute.

[0073] Furthermore, the first write amplification factor and the second write amplification factor obtained from the above retest are the current write amplification factors obtained by re-injecting data loads with different compression characteristics after pausing the composite aging load in step S3.

[0074] Specifically, when performance metrics decline, the test host first checks the fluctuation range of the first write amplification factor obtained from the retest relative to the first write amplification factor baseline, and the fluctuation range of the second write amplification factor obtained from the retest relative to the second write amplification factor baseline. If the fluctuation ranges of both write amplification factors are within the preset range, it indicates that the main controller of the tested solid-state drive has not significantly changed its data compression processing capability.

[0075] Based on this, if the temperature of the tested solid-state drive exceeds the preset temperature threshold and the media error count does not increase, it indicates that the performance degradation is not due to storage media damage or degradation of the main controller's compression function, but rather due to the main controller actively reducing its operating frequency because of excessive temperature. This is determined to be thermal throttling, which is the self-protection mechanism adopted by the tested solid-state drive when the temperature exceeds the safe operating range.

[0076] If the fluctuation range of both write amplification factors is within the preset range, and the temperature of the tested SSD does not exceed the preset temperature threshold, but the number of available spare blocks decreases or the media error count increases, it indicates that the NAND flash memory inside the tested SSD has experienced physical wear, which is determined to be an increase in bad blocks. The decrease in available spare blocks indicates that bad blocks have been replaced and have consumed spare resources, while the increase in the media error count indicates that the errors occurring in the storage cells during read and write operations are increasing.

[0077] If the difference between the first write amplification factor obtained from the retest and the second write amplification factor obtained from the retest exceeds a preset threshold relative to the difference between the baseline of the first write amplification factor and the baseline of the second write amplification factor, and the temperature of the tested solid-state drive does not exceed the preset temperature threshold and the media error count does not increase, it indicates that the main controller of the tested solid-state drive has degraded its ability to compress compressible data.

[0078] It is understandable that the difference between the first write amplification factor baseline and the second write amplification factor baseline reflects the compression processing capability of the main controller before the aging test, and the difference between the two write amplification factors obtained by retesting reflects the compression processing capability of the main controller after the aging test. If the change of the two exceeds the preset threshold, it indicates that the compression processing capability has decreased, and it is determined that the main controller compression engine has degraded.

[0079] It should be noted that the specific values ​​involved in the above-mentioned preset range, preset temperature threshold and preset threshold and other judgment conditions are parameters that can be flexibly set by those skilled in the art according to the actual application scenario such as the model and specifications of the solid-state drive under test and the type of NAND flash memory. This application does not impose any restrictions on them.

[0080] Step S5: When the anomaly level is a correctable anomaly level, perform the self-healing operation corresponding to the anomaly level, and continue testing after the self-healing operation is completed. In this embodiment of the application, after the test host determines the anomaly level corresponding to the current anomaly in step S4, it will take different processing methods according to the type of anomaly level.

[0081] When the current anomaly is determined to be a correctable anomaly, it means that the tested SSD has not yet suffered irrecoverable data corruption and there is still a possibility of restoring it to normal operation through intervention. At this time, the test host executes the self-healing operation corresponding to that anomaly level; different anomaly levels correspond to different self-healing operations. After the self-healing operation is completed, the test host continues to execute the composite aging load test process without manual intervention.

[0082] As an optional implementation method, please refer to Figure 5This is a flowchart of the self-healing operation corresponding to the first anomaly level in the solid-state drive closed-loop self-healing test method provided in this application embodiment. When the anomaly level is the first anomaly level, the step of executing the self-healing operation corresponding to the anomaly level includes: Step S500: Reduce the queue depth of the first load and extend the total test duration so that the total amount of data written by the composite aging load remains unchanged.

[0083] In an optional embodiment, when the anomaly level is the first anomaly level, the performance metrics of the tested SSD deviate from the performance baseline, but the media error count does not increase; the performance degradation is mainly due to excessive load pressure. At this time, the self-healing operation performed by the test host is to reduce the queue depth of the first load. After the queue depth of the first load is reduced, the number of read and write commands that the tested SSD needs to process per unit time decreases, and the processing pressure is reduced.

[0084] Meanwhile, the test host extended the total test duration to ensure that the total amount of data written to the tested SSD under composite aging load remained consistent throughout the entire test cycle compared to before the queue depth was reduced. By reducing instantaneous load pressure and extending the test duration to maintain a constant total write volume, performance stress can be alleviated without reducing aging stress.

[0085] As an optional implementation method, please refer to Figure 6 This is a flowchart of the self-healing operation corresponding to the second anomaly level in the solid-state drive closed-loop self-healing test method provided in this application embodiment. When the anomaly level is the second anomaly level, the step of executing the self-healing operation corresponding to the anomaly level includes: Step S510: Reduce the queue depth of the first load; Step S511: Perform a full disk read scan on the solid-state drive to trigger the background inspection and bad block management mechanism of the solid-state drive firmware; Step S512: After the scan is completed, restore the queue depth of the first load.

[0086] In an optional embodiment, when the anomaly level is the second anomaly level, the media error count has increased, but no correctable input / output error event carrying a logical address has been captured, indicating that there is an error in the storage medium but the location of the error has not yet been determined.

[0087] Specifically, the test host first reduces the queue depth of the initial load to decrease the write pressure on the SSD under test. Then, the test host performs a full disk read scan of the SSD under test, traversing and reading all logical address spaces.

[0088] During the full disk read scan, the firmware of the tested SSD will check the status of the physical storage unit corresponding to each logical address read. When it is found that the read data needs to be corrected or the read quality is lower than the threshold, the firmware will automatically trigger the background inspection and bad block management mechanism to repair or remap the physical storage unit with problems.

[0089] Furthermore, after the full disk scan is completed, the test host restores the queue depth of the first load to the level before the reduction and continues to execute the composite aging load test process.

[0090] As an optional implementation method, please refer to Figure 7 This is a flowchart of the self-healing operation corresponding to the third anomaly level in the solid-state drive closed-loop self-healing test method provided in this application embodiment. When the anomaly level is the third anomaly level, the step of executing the self-healing operation corresponding to the anomaly level includes: Step S520: Save the current test progress and load configuration status; Step S521: Issue a space reclamation command to the logical address range associated with the correctable input / output error event, so that the firmware of the solid-state drive reclaims the physical resources mapped by the logical address range and triggers bad block remapping processing. Step S522: After the space reclamation instruction is executed, resume the test based on the saved test progress and load configuration status.

[0091] In an optional embodiment, when the anomaly level is the third anomaly level, the test host has captured a correctable input / output error event carrying a logical address, and a perceptible error has occurred in the physical storage unit corresponding to a specific logical address range in the solid-state drive under test.

[0092] The test host will first save the current test progress and load configuration status. The test progress includes the current read and write positions that have been completed, and the load configuration status includes the current queue depth, read and write mode and other parameter settings of the first and second loads, so that the test can continue from the save point when it is restored in the future, avoiding retesting.

[0093] After saving the current test progress and load configuration status, the test host issues a space reclamation command to the logical address range associated with the correctable input / output error event. This space reclamation command is a type of command defined in standard storage protocols, used to notify the firmware of the SSD under test that the data within the specified logical address range is invalid, and the firmware can reclaim the physical resources mapped to that logical address range. Upon receiving the space reclamation command, the firmware marks the corresponding physical resources as reclaimable and allocates new physical storage units for the new data when subsequently rewriting to that logical address range.

[0094] During the reallocation process, the firmware evaluates the status of the originally mapped physical storage units. If the physical storage unit is found to be unreliable, it is marked as a bad block and a new physical storage unit is allocated from the spare blocks to replace it, thus completing the bad block remapping process.

[0095] Furthermore, after the space reclamation command is executed, the test host resumes the test based on the previously saved test progress and load configuration status. The recovery test refers to continuing the composite aging load from the saved test progress, restoring the load configuration parameters to the state at the time of saving, thereby completing the self-healing operation of the tested solid-state drive under the third anomaly level.

[0096] Step S6: When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

[0097] In this embodiment, when the current anomaly level is determined to be the fourth anomaly level (i.e., an uncorrectable anomaly level) in step S4, it indicates that the tested solid-state drive has experienced a data error that cannot be recovered by the internal error correction mechanism. The test host terminates the current composite aging test process and stops issuing new read / write commands to the tested solid-state drive.

[0098] Meanwhile, the test host records the logical address range where the uncorrectable error occurred. The logical address range refers to the range of logical addresses used by the test host when issuing read and write commands to the solid-state drive under test. Recording this logical address range provides location information for subsequent fault analysis and failure location, making it easier to determine which specific storage areas in the solid-state drive under test have suffered unrecoverable data corruption, thereby providing a basis for firmware optimization or hardware improvement of the solid-state drive under test.

[0099] Based on the above closed-loop self-healing test method for solid-state drives, please refer to... Figure 8 This is a functional block diagram of the solid-state drive closed-loop self-healing testing system provided in the embodiments of this application. This application also provides a solid-state drive closed-loop self-healing testing system, including: The acquisition unit is used to inject data payloads with different compression characteristics into the solid-state drive under test, collect the performance parameters and media operation parameters corresponding to each data payload, and construct an initial benchmark dataset that includes at least the write amplification factor baseline. The superposition unit is used to continuously apply a first load to the solid-state drive, and periodically superimpose a second load with a random access intensity higher than the first load without interrupting the first load, so as to form a composite aging load. The retesting unit is used to pause the composite aging load after completing a preset number of second load superpositions, and re-inject the data load with different compression characteristics into the solid-state drive under the same test conditions as when constructing the write amplification factor baseline to retest the current write amplification factor. The detection unit is used to collect the performance data and media health data of the solid-state drive in real time, and combine them with the initial benchmark dataset to perform anomaly detection and determine the anomaly level corresponding to the current anomaly. The processing unit is configured to, when the anomaly level is a correctable anomaly level, execute a self-healing operation corresponding to the anomaly level, and continue testing after the self-healing operation is completed; and, When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

[0100] For other details regarding the implementation of the above-mentioned solid-state drive closed-loop self-healing test system, please refer to the description of the solid-state drive closed-loop self-healing test method provided in the above-mentioned application embodiments, which will not be repeated here.

[0101] This application provides a closed-loop self-healing testing method and system for solid-state drives (SSDs). Before aging testing, data loads with different compression characteristics are injected into the SSD under test. Performance parameters and media operating parameters corresponding to each data load are collected to construct an initial benchmark dataset that includes at least a baseline write amplification factor. During the aging test phase, a first load is continuously applied to the SSD under test, and a second load with higher random access intensity is periodically superimposed without interrupting the first load, forming a composite aging load. The SSD under test is subjected to continuous steady-state read / write stress while being periodically impacted by high random access intensity loads, thus more closely resembling a real-world usage environment with sudden traffic surges.

[0102] After completing the second load stacking for a preset number of cycles, the composite aging load is paused. Under the same test conditions as when building the write amplification factor baseline, a data load with different compression characteristics is re-injected to retest the current write amplification factor. Real-time performance data and media health data of the tested SSD are collected and combined with the initial benchmark dataset for anomaly detection and anomaly level determination. When the anomaly level is correctable, the corresponding self-healing operation is executed, and testing continues after completion, preventing direct interruption due to minor anomalies and achieving automatic continuation of the test process. When the anomaly level is uncorrectable, the test is terminated, and the logical address range of the error is recorded, providing location information for subsequent fault analysis. This application can simulate the characteristics of sudden load stacking and execute corresponding self-healing operations according to the anomaly level, achieving automated testing while improving testing efficiency.

[0103] It should be noted that, in the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0104] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0105] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A closed-loop self-healing test method for solid-state drives, characterized in that, Includes the following steps: Data payloads with different compression characteristics are injected into the solid-state drive under test, and performance parameters and media operating parameters corresponding to each data payload are collected to construct an initial benchmark dataset that includes at least a write amplification factor baseline. The data payloads with different compression characteristics include a first data payload and a second data payload, where the first data payload is an incompressible data stream and the second data payload is a compressible data stream. The write amplification factor baseline in the initial benchmark dataset includes a first write amplification factor baseline corresponding to the first data payload and a second write amplification factor baseline corresponding to the second data payload, and the first write amplification factor baseline is different from the second write amplification factor baseline. A first load is continuously applied to the solid-state drive, and without interrupting the first load, a second load with a random access intensity higher than the first load is periodically superimposed to form a composite aging load. After completing the second load superposition for a preset number of times, the composite aging load is paused. Under the same test conditions as when constructing the write amplification factor baseline, the data load with different compression characteristics is re-injected into the solid-state drive to retest the current write amplification factor. Real-time collection of solid-state drive performance data and media health data, combined with the initial benchmark dataset for anomaly detection, to determine the anomaly level corresponding to the current anomaly; When the anomaly level is a correctable anomaly level, the self-healing operation corresponding to the anomaly level is executed, and the test continues after the self-healing is completed; wherein, when the performance index in the performance data deviates from the performance baseline value in the initial benchmark dataset, and the media error count in the media health data does not increase, the current anomaly is determined to be the first anomaly level among the correctable anomaly levels; and, when the anomaly level is the first anomaly level, the queue depth of the first load is reduced, and the total test duration is extended, so that the total amount of written data of the composite aging load remains unchanged; When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

2. The closed-loop self-healing test method for solid-state drives as described in claim 1, characterized in that, Determining the anomaly level corresponding to the current anomaly also includes: When the media error count in the media health data increases and no correctable input / output error event is captured during the execution of test read / write operations, it is determined to be at the second abnormal level; When a correctable input / output error event is captured during the execution of a test read / write operation, it is classified as a third-level anomaly. When an uncorrectable error status code is received from the solid-state drive, it is determined to be at the fourth level of abnormality. The second and third anomaly levels are correctable anomaly levels, while the fourth anomaly level is an uncorrectable anomaly level.

3. The closed-loop self-healing test method for solid-state drives as described in claim 2, characterized in that, When the anomaly level is the second anomaly level, the execution of the self-healing operation corresponding to the anomaly level includes: Reduce the queue depth of the first load; A full disk read scan is performed on the solid-state drive to trigger the background inspection and bad block management mechanism of the solid-state drive firmware; After the scan is complete, restore the queue depth of the first load.

4. The closed-loop self-healing test method for solid-state drives as described in claim 2, characterized in that, When the anomaly level is the third anomaly level, the execution of the self-healing operation corresponding to the anomaly level includes: Save the current test progress and load configuration status; A space reclamation command is issued to the logical address range associated with the correctable input / output error event, causing the firmware of the solid-state drive to reclaim the physical resources mapped by the logical address range and trigger bad block remapping processing. After the space reclamation command is executed, the test is resumed based on the saved test progress and load configuration status.

5. The solid-state drive closed-loop self-healing test method as described in claim 2, characterized in that, After determining the anomaly level corresponding to the current anomaly, the process also includes: When the performance metrics in the performance data decrease: If the fluctuation range of the first write amplification factor obtained by retesting relative to the baseline of the first write amplification factor and the fluctuation range of the second write amplification factor obtained by retesting relative to the baseline of the second write amplification factor are both within the preset range, the temperature of the solid-state drive exceeds the preset temperature threshold, and the media error count in the media health data does not increase, then the cause of the performance degradation of the solid-state drive is determined to be thermal throttling. If the fluctuation range of the first write amplification factor obtained by retesting relative to the baseline of the first write amplification factor and the fluctuation range of the second write amplification factor obtained by retesting relative to the baseline of the second write amplification factor are both within the preset range, the temperature of the solid-state drive does not exceed the preset temperature threshold, and the available spare blocks in the media health data decrease or the media error count increases, then it is determined that the cause of the performance degradation of the solid-state drive is the increase of bad blocks in the media. If the difference between the first write amplification factor obtained from the retest and the second write amplification factor obtained from the retest exceeds a preset threshold relative to the difference between the baseline of the first write amplification factor and the baseline of the second write amplification factor, and the temperature of the solid-state drive does not exceed a preset temperature threshold, and the media error count in the media health data does not increase, then the cause of the solid-state drive performance degradation is determined to be the degradation of the main control compression engine.

6. The closed-loop self-healing test method for solid-state drives as described in claim 1, characterized in that, After real-time acquisition of the solid-state drive's performance data and media health data, the method further includes: Based on the real-time collected performance data and media health data of the solid-state drive, the probability of failure within a future preset time window is obtained through time-series analysis. When the failure probability exceeds a preset probability threshold, the queue depth of the first load is proactively reduced before determining the anomaly level corresponding to the current anomaly.

7. The closed-loop self-healing test method for solid-state drives as described in claim 1, characterized in that, The queue depth of the second load is higher than the queue depth of the first load; and / or, The read / write addresses of the first load are distributed sequentially, while the read / write addresses of the second load are distributed randomly.

8. A closed-loop self-healing testing system for solid-state drives, characterized in that, include: The acquisition unit is used to inject data payloads with different compression characteristics into the solid-state drive under test, acquire the performance parameters and media operating parameters corresponding to each data payload, and construct an initial benchmark dataset that includes at least a write amplification factor baseline. The data payloads with different compression characteristics include a first data payload and a second data payload, where the first data payload is an incompressible data stream and the second data payload is a compressible data stream. The write amplification factor baseline in the initial benchmark dataset includes a first write amplification factor baseline corresponding to the first data payload and a second write amplification factor baseline corresponding to the second data payload, and the first write amplification factor baseline is different from the second write amplification factor baseline. The superposition unit is used to continuously apply a first load to the solid-state drive, and periodically superimpose a second load with a random access intensity higher than the first load without interrupting the first load, so as to form a composite aging load. The retesting unit is used to pause the composite aging load after completing a preset number of second load superpositions, and re-inject the data load with different compression characteristics into the solid-state drive under the same test conditions as when constructing the write amplification factor baseline to retest the current write amplification factor. The detection unit is used to collect the performance data and media health data of the solid-state drive in real time, and perform anomaly detection in combination with the initial benchmark dataset to determine the anomaly level corresponding to the current anomaly; wherein, when the performance index in the performance data deviates from the performance baseline value in the initial benchmark dataset, and the media error count in the media health data does not increase, the current anomaly is determined to be the first anomaly level among the correctable anomaly levels. The processing unit is configured to, when the anomaly level is a correctable anomaly level, execute a self-healing operation corresponding to the anomaly level, and continue testing after the self-healing operation is completed; wherein, when the anomaly level is a first anomaly level, the queue depth of the first load is reduced, and the total test duration is extended, so that the total amount of data written by the composite aging load remains unchanged; and, When the anomaly level is an uncorrectable anomaly level, terminate the test and record the logical address range where the error occurred.

Images