Performance testing method, device, storage medium, and program product

By populating the storage device with data and automatically switching test parameters according to set conditions, the problem of long test time caused by frequent start-stop of the pressure unit in the prior art is solved, realizing efficient and seamless performance testing and ensuring the continuity and accuracy of the test.

WO2026233000A1PCT designated stage Publication Date: 2026-11-12CLOUD INTELLIGENCE ASSETS HOLDING (SINGAPORE) PTE LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2026/085031
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-06
Filing Date
2026-03-20
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing technologies require frequent starting and stopping of the stress tester when performing performance tests on storage devices, resulting in long testing times, low efficiency, and an inability to effectively simulate sudden write scenarios during checkpoint saving.

Method used

After data is populated on the storage device, the test parameters for different pressure stages are automatically switched according to the set conditions to achieve seamless switching, avoid frequent start-stop of the pressure unit, and update the parameters using the switching conditions to ensure the continuity of the testing process.

Benefits of technology

It enables efficient performance testing of storage devices, reduces testing time, improves testing efficiency, accurately simulates real-world load scenarios, and avoids test interruptions and data loss.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2026085031_12112026_PF_FP_ABST
    Figure CN2026085031_12112026_PF_FP_ABST
Patent Text Reader

Abstract

The embodiments of the present disclosure relate to the technical field of computers, and provide a performance testing method, a device, a storage medium, and a program product. The method comprises: performing data population on a storage device to be tested; when the storage device to be tested satisfies a first set population condition, performing a data access operation on the storage device to be tested according to a first test parameter corresponding to a first stress phase; upon detecting that the data access operation corresponding to the first stress phase satisfies a first switching condition, performing the data access operation on the storage device to be tested according to a second test parameter corresponding to a second stress phase, the first test parameter being different from the second test parameter; and upon detecting that the data access operation corresponding to the second stress phase satisfies a second switching condition, continuing to perform the step of performing the data access operation according to the first test parameter corresponding to the first stress phase until the storage device to be tested satisfies a test termination requirement. The present solution allows seamless switching to be performed between different stress phases, consuming less time and achieving high testing efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Performance testing methods, equipment, storage media, and software products

[0001] This disclosure claims priority to Chinese Patent Application No. 202510582800.7, filed with the China Patent Office on May 6, 2025, entitled “Performance Testing Method, Apparatus, Storage Medium and Program Product”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This disclosure relates to the field of computer technology, and in particular to a performance testing method, apparatus, storage medium, and program product. Background Technology

[0003] During model training, checkpoints need to be saved periodically to ensure that training interruptions caused by hardware failures can be quickly recovered. During checkpoint saving, sudden writes may occur (i.e., large amounts of data are written to the storage device under test in a short period). In such cases, performance testing of the storage device becomes particularly important.

[0004] In existing technologies, performance testing of storage devices is typically performed using stress testers such as fio and recordstoreperf. In practice, the storage device is populated with data, and then a script continuously starts and stops the stress tester, alternating between high-pressure and low-pressure flows to simulate the data write characteristics of checkpoints (periodic bursts of data writes). However, this approach requires constant starting and stopping of the stress tester, and after each stop, test results need to be collected, summarized, and statistically analyzed before starting the next test, which is time-consuming. Summary of the Invention

[0005] This disclosure provides a performance testing method, device, storage medium, and program product that can seamlessly switch between different stress stages, consumes less time, and has high testing efficiency.

[0006] In a first aspect, embodiments of this disclosure provide a performance testing method, the method comprising:

[0007] Populate the storage device under test with data;

[0008] When the storage device under test meets the first set filling conditions, a data access operation is performed on the storage device under test according to the first test parameters corresponding to the first pressure stage.

[0009] If the data access operation corresponding to the first stress stage is detected to meet the first switching condition, the data access operation is performed on the storage device under test according to the second test parameters corresponding to the second stress stage, wherein the first test parameters and the second test parameters are different.

[0010] If the data access operation corresponding to the second stress stage is detected to meet the second switching condition, the step of executing the data access operation according to the first test parameters corresponding to the first stress stage continues until the storage device under test meets the test termination requirements.

[0011] Secondly, embodiments of this disclosure provide a performance testing apparatus, the apparatus comprising:

[0012] The filling module is used to fill the storage device under test with data.

[0013] The execution module is configured to perform a data access operation on the storage device under test according to a first test parameter corresponding to a first pressure stage when the storage device under test meets a first set filling condition, and to perform a data access operation on the storage device under test according to a second test parameter corresponding to a second pressure stage when the data access operation corresponding to the first pressure stage meets a first switching condition, wherein the first test parameter and the second test parameter are different.

[0014] The detection module is used to detect that the data access operation corresponding to the second pressure stage meets the second switching condition, and continue to execute the step of performing the data access operation according to the first test parameters corresponding to the first pressure stage until the storage device under test meets the test termination requirements.

[0015] Thirdly, embodiments of this disclosure provide a performance testing method, the method comprising:

[0016] Populate the storage device under test with data;

[0017] When the storage device under test meets the third preset filling condition, the first pressure device is invoked to perform a data access operation according to the first test parameters corresponding to the first pressure stage.

[0018] After the first pressure phase ends, the first pressure device is paused, and the second pressure device is invoked to perform data access operations according to the second test parameters corresponding to the second pressure phase.

[0019] After the second pressure phase ends, the invocation of the second pressure device is paused, and the step of invoking the first pressure device to perform data access operations according to the first test parameters corresponding to the first pressure phase continues until the storage device under test meets the test termination requirements.

[0020] Fourthly, embodiments of this disclosure provide a performance testing apparatus, the apparatus comprising:

[0021] The filling module is used to fill the storage device under test with data.

[0022] The calling module is configured to, when the storage device under test meets the third set filling condition, call the first pressure device to perform a data access operation according to the first test parameters corresponding to the first pressure stage, and after the first pressure stage ends, pause the calling of the first pressure device and call the second pressure device to perform a data access operation according to the second test parameters corresponding to the second pressure stage.

[0023] The execution module is configured to, after the second pressure phase ends, pause the invocation of the second pressure device and continue to execute the step of invoking the first pressure device to perform data access operations according to the first test parameters corresponding to the first pressure phase until the storage device under test meets the test termination requirements.

[0024] Fifthly, embodiments of this disclosure provide an electronic device, including: a memory, a processor, and a communication interface; wherein, the memory stores executable code, and when the executable code is executed by the processor, the processor performs the performance testing method as described in the first or third aspect.

[0025] In a sixth aspect, embodiments of this disclosure provide a non-transitory machine-readable storage medium storing executable code, which, when executed by a processor of an electronic device, enables the processor to at least implement the performance testing method as described in the first or third aspect.

[0026] In a seventh aspect, embodiments of this disclosure provide a computer program product, the computer program product including a computer program, which, when executed by a processor, is capable of implementing the performance testing method as described in the first or third aspect.

[0027] In the performance testing scheme provided in this embodiment, by filling the storage device under test with data, the actual required testing environment can be accurately simulated, such as the burst load environment saved during model training. After simulating the actual required testing environment, data access operations are performed on the storage device under test according to the first test parameters corresponding to the first stress stage. If the data access operation corresponding to the first stress stage is found to meet the first switching condition, data access operations are performed on the storage device under test according to the second test parameters corresponding to the second stress stage. During this process, if the data access operation corresponding to the second stress stage is found to meet the second switching condition, the step of performing data access operations according to the first test parameters corresponding to the first stress stage continues until the storage device under test meets the test termination requirements.

[0028] It should be noted that before the storage device under test meets the test termination requirements, the first and second stress phases will be continuously and alternately tested. Throughout the test, the switching between the first and second stress phases primarily relies on first and second switching conditions. When the first switching condition is met, the test parameters are directly updated, changing the first test parameter to the second test parameter, automatically switching from the first stress phase to the second stress phase. Similarly, when the second switching condition is met, the test parameters are also directly updated, changing the second test parameter to the first test parameter, automatically switching from the second stress phase to the first stress phase. In summary, this solution allows for seamless switching between the two stress phases by modifying the test parameters when the data access operations corresponding to different stress phases meet the switching conditions, without the need for frequent starting and stopping of the stress tester. Furthermore, after the test terminates, the final test results can be obtained uniformly. The entire testing process is uninterrupted, time-efficient, and highly effective. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1a is a schematic diagram of the application of a performance testing method provided by the prior art;

[0031] Figure 1b is a schematic diagram illustrating another application of a performance testing method provided by the prior art;

[0032] Figure 2 is a flowchart of a performance testing method provided in an embodiment of this disclosure;

[0033] Figure 3 is another flowchart of a performance testing method provided in an embodiment of this disclosure;

[0034] Figure 4 is a structural schematic diagram of a pressure device provided in an embodiment of this disclosure;

[0035] Figure 5 is a schematic diagram illustrating the application of a performance testing method provided in an embodiment of this disclosure;

[0036] Figure 6 is a specific example diagram of the target test results provided in the embodiments of this disclosure;

[0037] Figure 7 is a flowchart of another performance testing method provided in an embodiment of this disclosure;

[0038] Figure 8 is a structural schematic diagram of a performance testing device provided in an embodiment of this disclosure;

[0039] Figure 9 is another structural schematic diagram of a performance testing device provided in an embodiment of this disclosure;

[0040] Figure 10 is a schematic diagram of the structure of an electronic device provided in this embodiment. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure. In addition, the timing of the steps in the following method embodiments is only an example and not a strict limitation.

[0042] It should be noted that, in the cases involving user information in the embodiments of this disclosure, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this disclosure are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse. In addition, the various models involved in this disclosure (including but not limited to large language models or other models) comply with relevant laws and standards.

[0043] Currently, a script is typically used to continuously start and stop the pressure unit, alternately generating high-pressure and low-pressure flow rates to simulate the data write characteristics (periodic bursts of write flow) at checkpoints. For ease of understanding, the following explanation, in conjunction with Figures 1a and 1b, will provide a detailed explanation:

[0044] Figure 1a shows eight pressure stages, where H represents the high-pressure stage (high-pressure write flow stage) and L represents the low-pressure stage (low-pressure write flow stage). Each pressure stage lasts for a set time. By starting and stopping the pressure unit eight times, high-pressure and low-pressure flows are alternately generated to simulate the data write characteristics of the checkpoint. In practice, the storage device under test is first filled with data. For example, the storage capacity level of the storage device under test is raised to 80% of its total storage capacity to simulate a high-load situation. Then, the pressure unit is continuously started and stopped to alternately write data to the storage device under test during the high-pressure and low-pressure stages. During this process, after each pressure stage data writing is completed, the pressure unit needs to be shut down, and the results need to be statistically analyzed (such as the throughput changes of the high-pressure and low-pressure stages, the changes in the storage capacity level of the storage device under test, etc.) and data recovery needs to be performed (if the storage capacity level of the storage device under test rises due to data writing, the temporary data written to the storage device under test during the high-pressure and low-pressure stages is cleared to release the storage space of the storage device under test and restore the storage capacity level of the storage device under test). See Figure 1b for details. After that, the pressure unit is started again to conduct the next pressure stage test, which takes a long time.

[0045] In view of this, the present disclosure provides a performance testing method that addresses the above-mentioned problems based on the following approach: First, a first set filling condition is determined according to actual needs, and data is filled into the storage device under test according to the first set filling condition to accurately simulate the actual required test environment, such as the burst load environment of checkpoint saving during model training. Then, data access operations are alternately performed on the storage device under test according to the test parameters corresponding to the first and second stress stages. During the execution of each stress stage, it is checked in real time or at set intervals whether the stress stage switching condition has been met, and if it has, the stress stage is switched directly. This process is continuously iterated until the storage device under test meets the test termination requirements. It should be understood that the switching between different stress stages in the above test process is mainly based on the switching condition. When the switching condition is met, only internal modification of the test parameters is needed to achieve automatic and seamless switching between different stress stages. The entire test process does not require multiple restarts of the stress tester (i.e., the test process is never interrupted), consumes less time, and has high testing efficiency.

[0046] Figure 2 is a flowchart of a performance testing method provided in an embodiment of this disclosure. This method can be applied to a server (on which a stress tester can be running). As shown in Figure 2, the method includes the following steps:

[0047] 201. Populate the storage device under test with data.

[0048] 202. When the storage device under test meets the first set filling conditions, perform data access operations on the storage device under test according to the first test parameters corresponding to the first pressure stage.

[0049] 203. If the data access operation corresponding to the first stress stage is found to meet the first switching condition, the data access operation is performed on the storage device under test according to the second test parameters corresponding to the second stress stage. The first test parameters and the second test parameters are different.

[0050] 204. If the data access operation corresponding to the second stress stage is found to meet the second switching condition, continue to execute the steps of executing the data access operation according to the first test parameters corresponding to the first stress stage until the storage device under test meets the test termination requirements.

[0051] In practical applications, to meet actual performance testing needs, data can be filled into the storage device under test according to the first set filling conditions. Simply put, this involves writing data into the storage device under test to bring its storage capacity to the target level. For example, assuming a high-load environment needs to be simulated and the total storage capacity of the storage device under test is 10TB, data can be filled into the storage device under test according to the first set filling conditions (such as filling data into the storage device under test at a bandwidth of 5GB / s, filling the storage capacity to 80%), thus filling the storage space of the storage device under test to 8TB. ​​The storage device under test can be a hard disk drive (HDD), a solid-state drive (SSD), etc., and is not limited here.

[0052] If the storage device under test meets the aforementioned first set filling conditions, a target write request is sent to the storage device under test according to the first test parameters corresponding to the first stress stage. The first stress stage can be a high-pressure stage or a low-pressure stage. The first test parameters include the first request write duration, the first bandwidth, and the first rate limit. To facilitate understanding of "high pressure" and "low pressure," the following examples illustrate this: Assuming the first stress stage is a high-pressure stage, the first request write duration can be 5 minutes, the first bandwidth can be 5 GB / s, and the first rate limit can be 128. If the first stress stage is a low-pressure stage, the first request write duration can be 25 minutes (ensuring the storage device under test has sufficient time to complete garbage collection to maintain the water level), the first bandwidth can be 1 GB / s (approximately 20% of the high-pressure stage, avoiding excessive resource consumption), and the first rate limit can be 32 (reducing concurrent requests and lowering latency fluctuations). The first rate limit can also be referred to as the queue depth (QD).

[0053] For ease of description, the following explanation will use the example of the first pressure stage being the high-pressure stage and the second pressure stage being the low-pressure stage:

[0054] During the process of sending target write requests to the storage device under test according to the first test parameters corresponding to the first stress phase, the system checks in real time or at set intervals (e.g., 2s, 3s) whether the data access operation corresponding to the first stress phase meets the first switching condition. If it does not meet the condition, the system continues to send target write requests corresponding to the first test parameters to the storage device under test. If the condition is met, the system performs data access operations on the storage device under test according to the second test parameters corresponding to the second stress phase, that is, sends a setting request to the storage device under test. This setting request can be a write request, a data read request, a data delete request, etc., and is not limited here.

[0055] By sending target write requests to the storage device under test (DUT) according to the first test parameters corresponding to the first stress stage, under the condition that the first set filling conditions are met, high-pressure load simulation can be achieved. This allows the performance bottlenecks (such as garbage collection efficiency) of the DUT under high water levels to be exposed through continuous high-pressure writing. Furthermore, by determining whether the data access operations corresponding to the first stress stage meet the first switching conditions and processing accordingly based on the determination results, seamless switching between different stress stages can be achieved while ensuring the normal sending of target write requests.

[0056] During the data access operation performed on the storage device under test according to the second test parameters corresponding to the second stress phase, the system checks in real time or at set intervals whether the data access operation corresponding to the second stress phase meets the second switching condition. If it does not meet the condition, the system continues to perform data access operations on the storage device under test (such as sending write requests, data read requests, data delete requests, etc.). If the condition is met, the system returns to the steps of performing data access operations according to the first test parameters corresponding to the first stress phase and continues until the storage device under test meets the test termination requirements. The second test parameters include the second request write duration, the second bandwidth, and the second rate limit.

[0057] For example, regarding the execution process of performing data access operations on the storage device under test, the following is an example:

[0058] As one implementation, data access operations are performed on the storage device under test according to the first request write duration, first bandwidth, and first rate limit corresponding to the first stress phase. If the first request write duration is detected to be greater than the first duration threshold, it is determined that the data access operation corresponding to the first stress phase meets the first switching condition. Then, data access operations are performed on the storage device under test according to the second request write duration, second bandwidth, and second rate limit corresponding to the second stress phase. If the second request write duration is detected to be greater than the second duration threshold, it is determined that the data access operation corresponding to the second stress phase meets the second switching condition. At this point, the process returns to the step of performing data access operations according to the first test parameters corresponding to the first stress phase and continues until the storage device under test meets the test termination requirements.

[0059] In practical applications, for example, suppose the first stress phase corresponds to a first request write duration of 5 minutes, a first bandwidth of 5 GB / s, and a first rate limit of 128. The second stress phase corresponds to a second request write duration of 25 minutes, a second bandwidth of 1 GB / s, and a second rate limit of 32. Specifically, if the storage device under test (DUT) meets the first set filling conditions, data access operations are performed on the DUT according to the parameters of "first request write duration of 5 minutes, first bandwidth of 5 GB / s, and first rate limit of 128". During this process, if the first request write duration is detected to be greater than the first duration threshold (e.g., 4 minutes), it is determined that the data access operation corresponding to the first stress phase meets the first switching condition. At this time, data access operations are performed on the DUT according to the parameters of "second request write duration of 25 minutes, second bandwidth of 1 GB / s, and second rate limit of 32". During this process, if the second request write duration is detected to be greater than the second duration threshold (e.g., 24 minutes), the process returns to the step of performing data access operations according to the first test parameters corresponding to the first stress phase and continues until the DUT meets the test termination requirements.

[0060] By using the first request write duration and the second request write duration as the criteria for switching stress stages, the automatic switching between different stress stages can be accurately achieved, forming a periodic cycle. The entire process is automatic and requires no manual intervention, saving labor costs and increasing testing efficiency.

[0061] As another implementation, data access operations are performed on the storage device under test according to the first request write duration, first bandwidth, and first rate limit quantity corresponding to the first stress phase. If the storage level of the storage device under test is detected to be filled to the set threshold, it is determined that the data access operation corresponding to the first stress phase meets the first switching condition. Then, data access operations are performed on the storage device under test according to the second request write duration, second bandwidth, and second rate limit quantity corresponding to the second stress phase. If the garbage collection (GC) task of the storage device under test is detected to have been completed, it is determined that the data access operation corresponding to the second stress phase meets the second switching condition. At this time, the steps of performing data access operations according to the first test parameters corresponding to the first stress phase continue until the storage device under test meets the test termination requirements.

[0062] In practical applications, for example, suppose the first request write duration for the first stress phase is 5 minutes, the first bandwidth is 5 GB / s, and the first rate limit is 128. For the second stress phase, the second request write duration is 25 minutes, the second bandwidth is 1 GB / s, and the second rate limit is 32. Specifically, when the storage device under test meets the first set filling condition (filling the storage device under test with data at a bandwidth of 5 GB / s, so that the storage capacity of the storage device under test is filled to 80%), data access operations are performed on the storage device under test according to the following conditions: first request write duration of 5 minutes, first bandwidth of 5 GB / s, and first rate limit of 128. During this process, if the storage level of the storage device under test is detected to be filled to a set threshold (e.g., 85% of the total storage capacity of the storage device under test), exceeding the set threshold, then the data access operation corresponding to the first stress phase is determined to meet the first switching condition. At this time, the data access operation is performed on the storage device under test according to the following parameters: "second request write duration is 25 minutes, second bandwidth is 1GB / s, and second rate limit is 32". During this process, if the garbage collection task of the storage device under test is detected to be completed (indicating that the storage level of the storage device under test has recovered to below 80%), the process returns to the step of performing the data access operation according to the first test parameters corresponding to the first stress phase, continuing until the storage device under test meets the test termination requirements.

[0063] By using the storage level of the storage device under test as the first switching condition, we can more closely simulate the dynamic changes in storage space in real-world applications, avoiding test interruptions or data loss due to storage overflow. Furthermore, by using the execution status of the garbage collection task as the second switching condition, we can ensure that the storage device under test is in a state of sufficient resources to withstand the next round of stress testing, avoiding performance fluctuations caused by incomplete garbage collection tasks.

[0064] Furthermore, it is understood that when performing cyclic testing on the first and second pressure phases in the manner described above, a test termination requirement needs to be set to terminate the cyclic process. In this embodiment of the disclosure, the test termination requirement includes at least one of the following: the test duration reaches a set duration; the number of write tasks sent in different pressure phases reaches a set number threshold; and the number of cyclic iterations between the first and second pressure phases reaches a set number threshold.

[0065] To facilitate understanding, examples of the above test termination requirements are provided below:

[0066] Test termination requirement 1: The test duration reaches the set duration.

[0067] Assuming a test duration of one day, the performance tests of the first and second stress phases will be performed cyclically, automatically ending when the total test time reaches one day. By using the test duration as the criterion for test termination, sufficient test data from both the first and second stress phases can be ensured, while preventing the test from running indefinitely due to unexpected errors, infinite loops, or performance degradation, thus saving computational resources and time costs.

[0068] Test termination requirement 2: The number of write tasks sent during different pressure phases reaches the set threshold.

[0069] Assuming a threshold of 20,000 write tasks is set, the performance test will cycle through the first and second stress phases. The test will automatically terminate when the total number of write tasks in both phases reaches 20,000. By using the number of write tasks sent in different stress phases as the criterion for test termination, the test ensures that it covers sufficient data access operations, verifies the long-term stability of the storage device under test, and avoids overlooking potential problems due to insufficient testing.

[0070] Test termination requirement 3: The number of iterations between the first and second pressure phases reaches a set threshold.

[0071] Assuming a threshold of 100 iterations is set, the performance test will cycle through the first and second stress phases. The test will automatically terminate when the total number of iterations between the first and second stress phases reaches 100. By using the number of iterations between the first and second stress phases as the criterion for test termination, periodic service scenarios can be simulated, thereby verifying the stability of the storage device under test.

[0072] Test termination requirement 4: This can be any combination of the three test termination requirements mentioned above. For example, it can be that the test duration reaches the set duration and the number of write tasks sent in different pressure phases reaches a set threshold. Alternatively, it can be that the test duration reaches the set duration and the number of iterations between the first and second pressure phases reaches a set threshold, etc. These are not listed here.

[0073] The performance testing method provided in this disclosure can alternately perform data access operations on the storage device under test according to the test parameters corresponding to different stress stages. During the execution of each stress stage, it continuously detects whether the stress stage switching conditions have been met. If so, it directly switches to the next stress stage, and this process is continuously iterated until the storage device under test meets the test termination requirements. In summary, this disclosure embodiment can achieve seamless switching between two stress stages, with low time consumption and high testing efficiency.

[0074] Furthermore, after determining that the storage device under test meets the test termination requirements, the method further includes the following steps: obtaining the first test result corresponding to the first stress stage and the second test result corresponding to the second stress stage; filling the storage device under test with data; when the storage device under test meets the second set filling conditions, performing data access operations on the storage device under test according to the third test parameters, so that the write throughput of the storage device under test is within a predetermined range until the storage device under test meets the test termination requirements, so as to obtain the third test result, wherein the third test parameters are the same as or different from the first test parameters and the second test parameters, that is, the third test parameters can be the same as or different from the first test parameters, and can also be the same as or different from the second test parameters; and determining the target test result based on the first test result, the second test result, and the third test result.

[0075] In practical applications, we assume that the first pressure stage is the high-pressure stage and the second pressure stage is the low-pressure stage.

[0076] The test objective for the first stress phase can be: "Test the throughput of the storage device under test under extreme load", and the corresponding first test result can be: the throughput of the storage device under test under extreme load is 5GB / s, which meets the standard.

[0077] The test objective for the second stress phase can be: "Verify the resource recovery capability of the storage device under test under low load (such as verifying garbage collection efficiency)", and the corresponding second test result can be: garbage collection took 15 minutes, which meets the standard.

[0078] The test objective corresponding to "performing data access operations on the storage device under test according to the third test parameters to ensure that the write throughput of the storage device under test is within a predetermined range" is to test the performance changes of the storage device under test under long-term high load and to verify the stability of the storage device under test under continuous constant voltage. Specifically, assuming that the third test parameters include: third bandwidth (assumed to be 3GB / s), third rate limit (assumed to be 64), and third request write duration (assumed to be 24 hours), then in practical applications, if data access operations are performed on the storage device under test according to the above third test parameters, the third test result can be obtained. This third test result can be "average throughput of 2.9GB / s, stable throughput, no other significant performance fluctuations, and the storage device under test is operating normally." In this process, the second set filling condition can be the same as or different from the first set filling condition, which can be determined according to the actual situation and is not limited here. The test termination requirements in this process can be: whether the total test duration reaches the set duration threshold, whether the cumulative amount of data written to the storage device under test reaches the set data amount threshold, etc., which are not listed here.

[0079] After obtaining the first test result, the second test result, and the third test result, the target test result can be determined based on the first test result, the second test result, and the third test result.

[0080] As one approach, the first, second, and third test results can be directly integrated and analyzed to obtain the final target test result. Continuing with the example of the first, second, and third test results, the target test result formed after integration and analysis can be "The storage device under test performs well in the high-voltage stage, performs excellently in the low-voltage stage, and has good stability under constant voltage."

[0081] As another implementation, different weights can be configured for the first, second, and third test results. For example, the weights of the first and second test results can both be configured to 0.4, while the weight of the third test result can be configured to 0.2. Then, based on the examples of the first, second, and third test results, scores are assigned to each result. For instance, if the first test result is 80 points, the second test result is 100 points, and the third test result is 90 points, then the target test result score can be 80 × 0.4 + 100 × 0.4 + 90 × 0.2 = 90 points, indicating that the overall test result of the storage device under test is excellent.

[0082] By obtaining the first, second, and third test results, and determining the final target test result based on these results, the reference factors for the target test result become more diverse, and the coverage of scenarios becomes more comprehensive (such as sudden load scenarios, daily load scenarios, long-term stable scenarios, etc.), thus ensuring the accuracy of the test results.

[0083] Figure 3 is another flowchart of a performance testing method provided in an embodiment of this disclosure. As shown in Figure 3, the method includes the following steps:

[0084] 301. Populate the storage device under test with data.

[0085] 302. When the storage device under test meets the first set filling conditions, send a target write request to the storage device under test according to the first test parameters corresponding to the first pressure stage.

[0086] 303. After any target write request response is completed, collect at least one performance metric data of the storage device under test.

[0087] 304. Adjust the first test parameter based on at least one performance indicator data to continue sending target write requests to the storage device under test using the adjusted first test parameter.

[0088] 305. If the data access operation corresponding to the first pressure stage is found to meet the first switching condition, the data access operation is executed on the storage device under test according to the second test parameters corresponding to the second pressure stage.

[0089] 306. If the data access operation corresponding to the second stress stage is found to meet the second switching condition, continue to execute the steps of executing the data access operation according to the first test parameters corresponding to the first stress stage until the storage device under test meets the test termination requirements.

[0090] The specific execution process of steps 301-302 and steps 305-306 can be found in the above embodiments, and will not be repeated here.

[0091] For steps 303-304, examples are provided below for ease of understanding:

[0092] Assume the first pressure phase is the low-pressure phase (the first write request takes 25 minutes). During the low-pressure phase, a total of 10 target write requests are sent to the storage device under test. After the 6th target write request is completed, a callback process is performed to collect the performance metrics data (such as latency, throughput, etc.) corresponding to the 6th target write request, as well as the overall throughput, average latency, storage water level, etc. from the 1st to the 6th target write request.

[0093] In practical applications, assuming target write requests 1-5 each write 1GB of data to the storage device under test, with a stable latency of 50ms, the cumulative throughput is 5GB / 5 seconds = 1GB / s. However, after target write request 6 writes 1GB of data to the storage device under test, the latency suddenly increases to 150ms (possibly because the storage device under test starts performing garbage collection).

[0094] After statistically analyzing at least one performance metric of the aforementioned storage device under test, it was found that: average throughput = 6GB / 6 seconds = 1GB / s, and average latency = (5×50+150) / 6 ≈ 66.7ms. Therefore, the concurrent target write requests can be reduced by decreasing the first rate limit (i.e., queue depth) to alleviate the garbage collection pressure on the storage device. For example, reducing the first rate limit from 32 to 16 will allow subsequent target write requests 7 through 10 to be sent to the storage device under test with a first rate limit of 16. Furthermore, assuming the storage capacity of the storage device under test changes from 80% to 80.6%, the write time for the first request can be increased from 25 minutes to 30 minutes to ensure successful garbage collection.

[0095] In summary, by collecting at least one performance metric data of the storage device under test after any target write request response is completed, and adjusting the first test parameter based on the at least one performance metric data, the system can continue to send target write requests to the storage device under test using the adjusted first test parameter. This achieves dynamic adjustment of the first test parameter during the first stress test phase. The entire process does not require restarting the stress tester, avoiding test interruption and ensuring the successful completion of the first stress test phase.

[0096] As another implementation method, after any target write request is completed, a callback process is performed to collect and update at least one performance indicator data of the storage device under test. For example, the number of requests sent is 15, the current target write request takes 50ms, the current cumulative written data is 25GB, the current throughput is 5GB / s, the current storage capacity of the storage device under test is 87%, etc. The first request write time in the first stress phase is 5 minutes, which will not be explained in detail here.

[0097] Based on at least one performance metric of the storage device under test, it can be determined whether the switching conditions for the stress phase have been met, or whether to continue sending new target write requests to the storage device under test.

[0098] Specifically, assuming the first pressure phase is a high-pressure phase, current statistics show that the storage capacity level of the tested storage device is 87%, which is greater than the set threshold (e.g., 85%). Therefore, the data access operation corresponding to the first pressure phase meets the first switching condition, and the pressure phase can be switched. Additionally, statistics show that the first request write duration in the first pressure phase is 5 minutes, reaching the first set duration threshold (e.g., 5 minutes). At this point, the pressure phase can also be switched.

[0099] By performing callback processing after any target write request response is completed, it can be determined whether the switching conditions for the stress phase are met. It should be noted that the embodiments of this disclosure not only support periodic switching (such as switching when the first request write duration reaches a first set duration threshold), but also support abnormal response switching (such as when the storage level of the storage device under test is greater than a set level threshold), ensuring the accuracy of stress phase switching.

[0100] In addition, when the storage device under test meets the first set filling condition, a target write request is sent to the storage device under test according to the first rate limit number corresponding to the first pressure stage. If it is detected that the data access operation corresponding to the first pressure stage does not meet the first switching condition, the number of requests sent is counted to see if the first rate limit number has been reached. If the number of requests sent has not reached the first rate limit number, the target write request corresponding to the first test parameter is sent to the storage device under test.

[0101] For example, assuming the first stress phase is a high-pressure phase with a corresponding first current limit of 128, then, if the storage device under test (DUT) meets the first set filling condition, target write requests are sent to the DUT according to the first current limit of 128. During this process, if it is detected that the data access operation corresponding to the first stress phase does not meet the first switching condition, the number of sent requests is checked to see if it has reached the first current limit. If the number of sent requests has not reached the first current limit, target write requests continue to be sent to the DUT. It can be understood that if the number of sent requests has not reached the first current limit, target write requests can continue to be sent to the DUT to fully utilize the storage device's throughput capacity and quickly fill the request queue.

[0102] If the number of sent requests reaches the first rate limit, the sending of target write requests to the storage device under test is suspended; after any first request response is completed, the number of sent requests is updated; if the number of sent requests is detected to be less than the first rate limit, the sending of target write requests to the storage device under test continues.

[0103] In practical implementation, for example, suppose the number of requests sent is 128, which has reached the first rate limit. At this time, the sending of target write requests to the storage device under test is paused. After any first request response is completed, a callback is performed and the number of requests sent is updated. It should be understood that the number of requests sent may change after any first request response is completed. If the current number of requests sent changes from 128 to 127, it means that the number of requests sent is less than the first rate limit. At this time, the sending of target write requests to the storage device under test can continue.

[0104] By pausing the sending of target write requests to the storage device under test (DUT) when the number of sent requests reaches the first rate limit, overload of the DUT can be avoided, ensuring the stability of the DUT's handling of target write requests. Conversely, by continuing to send target write requests to the DUT when the number of sent requests is detected to be less than the first rate limit, the throughput capacity of the storage device can be fully utilized while reducing request backlog within the DUT.

[0105] To facilitate understanding, the following example illustrates this solution in detail using a pressure regulator as an example, combined with the checkpoint writing scenario during model training:

[0106] Figure 4 is a schematic diagram of a pressure device provided in an embodiment of this disclosure. As shown in Figure 4, the pressure device includes multiple target tasks (such as checkpoint writing tasks, data recovery tasks, etc.), and each target task corresponds to multiple modules. These multiple modules include: a chunktable module, a generator module, a read / write management module (also known as an IO management module), a water level management module, a power-on recovery module, a real-time statistics module, and a current limiting module.

[0107] The block table module is used to manage the metadata of data blocks (such as the location, size, and status of the data blocks). It should be noted that the checkpoint data is pre-divided into multiple fixed-size data blocks (such as 1GB), and the block table module records the storage location of each data block in the storage device under test.

[0108] The generator module is used to generate test data or real-world application data (such as the contents of checkpoints). For example, in a specific implementation, the generator module can generate random data at a speed of 5GB / s to simulate the data writing to the storage device under test during the model training process.

[0109] The IO management module controls data read and write operations, including request sending, queue management, and error handling. For example, during high-pressure phases, the IO management module can send write requests to the storage device under test with a rate-limited limit of 128, maximizing the throughput of the storage device.

[0110] The water level management module is used to monitor the storage capacity level of the storage device under test. In practice, if the water level management module detects that the current storage capacity level of the storage device under test has reached the set water level threshold, it can trigger the deletion of old data.

[0111] The power-on recovery module is used to repair incomplete write operations by replaying logs after the pressure unit is powered off and restarted.

[0112] The real-time statistics module is used to collect and display the performance indicators of the storage device under test during the test, such as throughput, latency, and storage capacity.

[0113] The current limiting module is used to control the amount of data flow generated by the pressure unit, and supports the constant current limiting strategy (setting the pressure throughput to a constant value to generate a stable pressure flow) and the periodic current limiting strategy (simulating periodically changing pressure throughput by alternating between testing the first pressure stage and the second pressure stage).

[0114] It's important to note that during the checkpoint write scenario in model training, the model is trained on the Graphics Processing Unit (GPU). During checkpoint writes, training pauses, and the GPU copies the model state (such as parameters, optimizer state, and training progress) from the GPU to host memory. Then, the single-machine storage engine serializes the data in host memory and saves it to the storage device under test. During this process, the GPU must wait until the data is completely written to the storage device under test before resuming training. The single-machine storage engine, as a software-level storage management module, is responsible for controlling data read / write logic, garbage collection, etc., and its performance depends on the physical capabilities of the storage device under test (such as an SSD). The throughput, latency, and GC efficiency of the storage device under test, as the physical storage medium, directly affect the performance of the storage engine. Therefore, the performance test conducted in this solution can be understood as a collaborative performance evaluation of the single-machine storage engine and the storage device under test, aiming to verify their combined performance under high load scenarios.

[0115] In practice, as shown in Figure 5, the test process includes the following steps:

[0116] S1. Generate the target task based on multiple checkpoints generated during model training.

[0117] The target task includes a first set filling condition (also known as the target water level), a first test parameter corresponding to the first pressure stage, and a second test parameter corresponding to the second pressure stage. The target task can be target task 1 in Figure 4 (such as a checkpoint write task). Under this task, the target water level (such as 80% of the storage capacity of the storage device under test), the first test parameter corresponding to the first pressure stage (such as a first bandwidth of 40GB / s, a first request write duration of 300s, etc.), and the second test parameter corresponding to the second pressure stage (such as a second bandwidth of 100MB / s, a second request write duration of 600s, etc.) are defined.

[0118] S2. Execute the target task to fill the storage device under test with data according to the target water level.

[0119] By generating and executing target tasks based on multiple checkpoints generated during model training, and filling the storage device under test with data according to the target water level, it is possible to simulate a high-load environment and lay the foundation for subsequent performance testing.

[0120] S3. Send a target write request to the storage device under test according to the first test parameters corresponding to the first pressure stage.

[0121] S4. After any target write request response is completed, a callback process is performed.

[0122] S5. Based on the callback processing result, determine whether the data access operation corresponding to the first pressure stage meets the first switching condition. If it does, execute S6; otherwise, execute S7.

[0123] S6. Perform data access operations on the storage device under test according to the second test parameters corresponding to the second pressure stage.

[0124] S7. Determine if the number of requests sent has reached the first rate limit. If it has, proceed to S8; otherwise, proceed to S9.

[0125] S8. Pause sending target write requests to the storage device under test and wait for a set time. When it is determined that the number of requests sent has not reached the first rate limit, execute S9.

[0126] S9. Continue sending target write requests to the storage device under test.

[0127] S10. If the data access operation corresponding to the second pressure stage is detected to meet the second switching condition, return to the step of executing the data access operation according to the first test parameters corresponding to the first pressure stage and continue to execute until the storage device under test meets the test termination requirements and the test ends.

[0128] It should be noted that a target test result will be generated after the test, and a specific example of this target test result can be seen in Figure 6. In Figure 6, the storage device under test performed well in the first three high-pressure stages. However, in the subsequent high-pressure stages, the single-disk maximum write throughput decreased by approximately 50%. Analysis suggests that this decrease in throughput may be due to increased garbage collection within the disk. Furthermore, as shown in Figure 6, the transition between the high-pressure and low-pressure stages was smooth, achieving seamless switching between the two stress stages and ensuring overall performance testing efficiency.

[0129] Figure 7 is a flowchart of another performance testing method provided in this embodiment of the present disclosure. The executing entity of this method can be a terminal device with control functions. As shown in Figure 7, the method includes the following steps:

[0130] 701. Populate the storage device under test with data.

[0131] 702. When the storage device under test meets the third set filling conditions, the first pressure device is invoked to perform data access operation according to the first test parameters corresponding to the first pressure stage.

[0132] 703. After the first pressure stage ends, pause the invocation of the first pressure unit and invoke the second pressure unit to perform data access operations according to the second test parameters corresponding to the second pressure stage.

[0133] 704. After the second pressure phase ends, pause the invocation of the second pressure unit and continue to execute the steps of invoking the first pressure unit to perform data access operations according to the first test parameters corresponding to the first pressure phase until the storage device under test meets the test termination requirements.

[0134] In practical applications, to meet actual performance testing needs, data can be filled into the storage device under test according to a third set filling condition (which may be the same as or different from the first set filling condition). Simply put, this involves writing data into the storage device under test to bring its storage capacity to the target level. For example, assuming the goal is to simulate a high-load environment and the total storage capacity of the storage device under test is 10TB, data can be filled into the storage device under test according to the first set filling condition (e.g., filling data into the storage device under test at a bandwidth of 5GB / s, filling the storage capacity to 80%), thus filling the storage space to 8TB. ​​The storage device under test can be a hard disk drive (HDD), a solid state drive (SSD), etc.

[0135] When the storage device under test meets the above-mentioned filling conditions, the first pressure unit is invoked to send a target write request to the storage device under test according to the first test parameters corresponding to the first pressure stage. The first pressure stage can be a high-pressure stage or a low-pressure stage. The first test parameters include the first request write duration, the first bandwidth, and the first rate limit. To facilitate understanding of "high pressure" and "low pressure," the following examples illustrate this: Assuming the first pressure stage is a high-pressure stage, the first request write duration can be 5 minutes, the first bandwidth can be 5 GB / s, and the first rate limit can be 128. If the first pressure stage is a low-pressure stage, the first request write duration can be 25 minutes (ensuring the storage device under test has sufficient time to complete garbage collection to maintain the water level), the first bandwidth can be 1 GB / s (approximately 20% of the high-pressure stage, avoiding excessive resource consumption), and the first rate limit can be 32 (reducing concurrent requests and lowering latency fluctuations). The first rate limit can also be referred to as the queue depth (QD).

[0136] After the first stress phase ends, the invocation of the first stressor is paused, and the second stressor is invoked to perform data access operations according to the second test parameters corresponding to the second stress phase. After the second stress phase ends, the invocation of the second stressor is paused, and the process of invoking the first stressor to perform data access operations according to the first test parameters corresponding to the first stress phase continues until the storage device under test meets the test termination requirements. The specific execution process of this part can be found in the above embodiment, and will not be repeated here.

[0137] By using two pressure units to switch between the high-pressure and low-pressure stages, the problem of long testing time and low efficiency caused by the continuous starting and stopping of a single pressure unit can be avoided. In addition, the modification of a single pressure unit can be reduced. That is, there is no need to add the current limiting module in the above embodiment to a single pressure unit. Only an interface with pause and activation functions needs to be added, which saves the modification cost of a single pressure unit.

[0138] The performance testing apparatus of one or more embodiments of this disclosure will be described in detail below. Those skilled in the art will understand that these apparatuses can be configured using commercially available hardware components through the steps taught in this solution.

[0139] Figure 8 is a schematic diagram of a performance testing device provided in an embodiment of this disclosure. As shown in Figure 8, the device includes a filling module 81, an execution module 82, and a detection module 83.

[0140] The filling module 81 fills the storage device under test with data.

[0141] The execution module 82 is configured to perform a data access operation on the storage device under test according to a first test parameter corresponding to a first pressure stage when the storage device under test meets a first set filling condition, and to perform a data access operation on the storage device under test according to a second test parameter corresponding to a second pressure stage when the data access operation corresponding to the first pressure stage meets a first switching condition, wherein the first test parameter and the second test parameter are different.

[0142] The detection module 83 is used to detect that the data access operation corresponding to the second pressure stage meets the second switching condition, and continue to execute the step of executing the data access operation according to the first test parameters corresponding to the first pressure stage until the storage device under test meets the test termination requirements.

[0143] Optionally, the execution module 82 is specifically used to: when the storage device under test meets the first set filling condition, send a target write request to the storage device under test according to the first test parameters.

[0144] The device further includes: a sending module, configured to continue sending a target write request corresponding to the first test parameter to the storage device under test when the data access operation corresponding to the first pressure stage is detected to not meet the first switching condition; and to collect at least one performance index data of the storage device under test after any target write request response is completed; and to adjust the first test parameter according to the at least one performance index data so as to continue sending target write requests to the storage device under test through the adjusted first test parameter.

[0145] Optionally, the sending module is specifically configured to: when the storage device under test meets the first set filling condition, send a target write request to the storage device under test according to the first rate limit; and when it is detected that the data access operation corresponding to the first pressure stage does not meet the first switching condition, count whether the number of sent requests has reached the first rate limit; and when the number of sent requests has not reached the first rate limit, continue to send a target write request corresponding to the first test parameter to the storage device under test.

[0146] Optionally, the sending module is further configured to: pause sending target write requests to the storage device under test when the number of sent requests reaches the first rate limit; update the number of sent requests after any first request response is completed; and if it is detected that the number of sent requests has not reached the first rate limit, continue sending target write requests corresponding to the first test parameters to the storage device under test.

[0147] Optionally, the filling module 81 is specifically used to: generate a target task based on multiple checkpoints generated during model training; the target task includes the first set filling conditions, the first test parameters corresponding to the first pressure stage, and the second test parameters corresponding to the second pressure stage; and execute the target task to fill the storage device under test with data according to the target water level.

[0148] Optionally, the first test parameters include a first request write duration, a first bandwidth, and a first rate limit; the second test parameters include a second request write duration, a second bandwidth, and a second rate limit; the execution module 82 is specifically used for:

[0149] If the first request write duration is detected to be greater than the first duration threshold, then it is determined that the data access operation corresponding to the first stress phase meets the first switching condition. Data access operations are then performed on the storage device under test according to the second request write duration, the second bandwidth, and the second rate limiting quantity corresponding to the second stress phase. The step of detecting that the data access operation corresponding to the second stress phase meets the second switching condition includes: if the second request write duration is detected to be greater than the second duration threshold, then it is determined that the data access operation corresponding to the second stress phase meets the second switching condition.

[0150] Optionally, the apparatus further includes: a determining module, configured to acquire a first test result corresponding to the first pressure stage and a second test result corresponding to the second pressure stage; fill the storage device under test with data according to a second preset filling condition; when the storage device under test meets the second preset filling condition, perform a data access operation on the storage device under test according to a third test parameter, so that the write throughput of the storage device under test is within a predetermined range until the storage device under test meets the test termination requirement, to obtain a third test result, wherein the third test parameter is the same as or different from the first test parameter and the second test parameter; and determine a target test result based on the first test result, the second test result, and the third test result.

[0151] Optionally, the test termination requirements include at least one of the following: the test duration reaches a set duration; the number of write tasks sent in different pressure phases reaches a set number threshold; and the number of iterations between the first pressure phase and the second pressure phase reaches a set number threshold.

[0152] The device shown in Figure 8 can perform the steps in the performance testing method in the foregoing embodiments. For detailed execution process and technical effects, please refer to the description in the foregoing embodiments, which will not be repeated here.

[0153] Figure 9 is a structural schematic diagram of a performance testing device provided in an embodiment of this disclosure. As shown in Figure 8, the device includes: a filling module 91, a calling module 92, and an execution module 83.

[0154] The filling module 91 is used to fill the storage device under test with data.

[0155] The calling module 92 is used to call the first pressure device to perform data access operations according to the first test parameters corresponding to the first pressure stage when the storage device under test meets the third set filling conditions, and to pause the calling of the first pressure device and call the second pressure device to perform data access operations according to the second test parameters corresponding to the second pressure stage after the first pressure stage ends.

[0156] The execution module 93 is used to pause the invocation of the second pressure device after the second pressure stage ends, and continue to execute the step of invoking the first pressure device to perform data access operations according to the first test parameters corresponding to the first pressure stage until the storage device under test meets the test termination requirements.

[0157] The device shown in Figure 9 can perform the steps in the performance testing method in the foregoing embodiments. For detailed execution process and technical effects, please refer to the description in the foregoing embodiments, which will not be repeated here.

[0158] Figure 10 is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. As shown in Figure 10, in practice, this electronic device includes a memory 101 and a processor 102.

[0159] Memory 101 is used to store computer programs and can be configured to store various other data to support operation on the electronic device. Examples of this data include instructions for any application or method used to operate on the electronic device, data structures, contact data, phone book data, messages, pictures, videos, etc.

[0160] The processor 102, coupled to the memory 101, is used to execute the computer program in the memory 101 to implement the performance testing method provided in the foregoing embodiments.

[0161] Furthermore, as shown in Figure 10, the electronic device also includes other components such as a communication component 103, a display 104, a power supply component 105, and an audio component 106. Figure 10 only schematically shows some components and does not imply that the electronic device includes only the components shown in Figure 10. The electronic device of this embodiment can be a terminal device such as a desktop computer, laptop computer, smartphone, or IoT device, or a server device such as a conventional server, cloud server, or server array.

[0162] The aforementioned memory can be implemented by any type of volatile or non-volatile storage device under test or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0163] The aforementioned communication component is configured to facilitate wired or wireless communication between the device containing the communication component and other devices. The device containing the communication component can access wireless networks based on communication standards, such as 2G, 3G, 4G / LTE, 5G, or combinations thereof. In one exemplary embodiment, the communication component receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel.

[0164] The aforementioned display includes a screen, which may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a Touch Panel, the screen can be implemented as a touchscreen to receive input signals from the user. The Touch Panel includes one or more touch sensors to sense touches, swipes, and gestures on the Touch Panel. The touch sensors can sense not only the boundaries of touch or swipe actions but also the duration of write requests and pressure associated with the touch or swipe operation.

[0165] The aforementioned power supply components provide power to various components within the device in which they reside. These power supply components may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device in which they reside.

[0166] The aforementioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC) configured to receive external audio signals when the device containing the audio component is in an operating mode, such as call mode, recording mode, or voice recognition mode. The received audio signals can be further stored in memory or transmitted via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0167] Accordingly, this disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the steps in the above-described method embodiments. The computer-readable storage medium includes volatile or non-volatile or a combination thereof, and can be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technologies, CD-ROM, Digital Video Disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices under test, or any other non-transfer medium.

[0168] Accordingly, this disclosure also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, cause the processor to implement the steps in the above-described method embodiments. It should be understood that each step or combination of steps in the above-described method flow can be implemented by the computer program or instructions. Furthermore, these computer programs or instructions can be applied to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device, enabling the processor of the general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to function as an apparatus for implementing the corresponding functions in the above-described method embodiments.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit them. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A performance testing method, wherein, include: Populate the storage device under test with data; When the storage device under test meets the first set filling conditions, a data access operation is performed on the storage device under test according to the first test parameters corresponding to the first pressure stage. If the data access operation corresponding to the first stress stage is detected to meet the first switching condition, the data access operation is performed on the storage device under test according to the second test parameters corresponding to the second stress stage, wherein the first test parameters and the second test parameters are different. If the data access operation corresponding to the second stress stage is detected to meet the second switching condition, the step of executing the data access operation according to the first test parameters corresponding to the first stress stage continues until the storage device under test meets the test termination requirements.

2. The method according to claim 1, wherein, When the storage device under test meets the first set filling condition, performing a data access operation on the storage device under test according to the first test parameters corresponding to the first pressure stage includes: If the storage device under test meets the first set filling condition, a target write request is sent to the storage device under test according to the first test parameters; The method further includes: If the data access operation corresponding to the first stress phase is found to fail to meet the first switching condition, the target write request corresponding to the first test parameter is sent to the storage device under test.

3. The method according to claim 2, wherein, Also includes: After any target write request response is completed, at least one performance metric of the storage device under test is collected. The first test parameter is adjusted based on the at least one performance metric data, so that the target write request can continue to be sent to the storage device under test using the adjusted first test parameter.

4. The method according to claim 2, wherein, The first test parameter includes a first rate limit; the step of sending a target write request to the storage device under test according to the first test parameter when the storage device under test meets the first set fill condition includes: If the storage device under test meets the first set filling condition, a target write request is sent to the storage device under test according to the first current limit quantity; The step of continuing to send a target write request corresponding to the first test parameter to the storage device under test when the data access operation corresponding to the first pressure stage is detected as not meeting the first switching condition includes: If the data access operation corresponding to the first pressure phase is not satisfied with the first switching condition, count whether the number of sent requests has reached the first rate limit. If the number of requests sent has not reached the first rate limit, continue to send target write requests corresponding to the first test parameters to the storage device under test.

5. The method according to claim 4, wherein, After determining whether the number of sent requests has reached the first rate limit, the method further includes: If the number of sent requests reaches the first rate limit, the sending of target write requests to the storage device under test will be suspended. After any first request response is completed, update the number of requests sent; If it is detected that the number of requests sent has not reached the first rate limit, then continue to send target write requests corresponding to the first test parameters to the storage device under test.

6. The method according to any one of claims 1-5, wherein, The data filling process for the storage device under test includes: Based on multiple checkpoints generated during model training, a target task is generated; the target task includes the first set filling condition, the first test parameter corresponding to the first stress stage, and the second test parameter corresponding to the second stress stage. The target task is to be performed to populate the storage device under test with data according to the target water level.

7. The method according to any one of claims 1-5, wherein, The first test parameters include the first request write duration, the first bandwidth, and the first rate limit; the second test parameters include the second request write duration, the second bandwidth, and the second rate limit. If the data access operation corresponding to the first stress phase is detected to meet the first switching condition, then performing a data access operation on the storage device under test according to the second test parameters corresponding to the second stress phase includes: If the first request write duration is detected to be greater than the first duration threshold, it is determined that the first stress phase meets the first switching condition, and data access operations are performed on the storage device under test according to the second request write duration, the second bandwidth, and the second rate limit. The condition that the data access operation corresponding to the second pressure stage meets the second switching condition includes: If the write duration of the second request is detected to be greater than the second duration threshold, then it is determined that the data access operation corresponding to the second stress phase meets the second switching condition.

8. The method according to any one of claims 1-5, wherein, The method further includes the step of continuing to execute the data access operation according to the first test parameters corresponding to the first stress stage after the data access operation corresponding to the second stress stage is detected to meet the second switching condition, until the storage device under test meets the test termination requirements. Obtain the first test result corresponding to the first pressure stage and the second test result corresponding to the second pressure stage; Data is populated into the storage device under test; When the storage device under test meets the second set filling condition, data access operation is performed on the storage device under test according to the third test parameters so that the write throughput of the storage device under test is within a predetermined range until the storage device under test meets the test termination requirement, so as to obtain the third test result. The third test parameters are the same as or different from the first test parameters and the second test parameters. Based on the first test result, the second test result, and the third test result, the target test result is determined.

9. The method according to any one of claims 1-5, wherein, The test termination requirement includes at least one of the following: the test duration reaches the set duration; The number of write tasks sent during different pressure phases reaches a set threshold; and the number of iterations between the first pressure phase and the second pressure phase reaches a set threshold.

10. A performance testing method, wherein, include: Populate the storage device under test with data; When the storage device under test meets the third preset filling condition, the first pressure device is invoked to perform a data access operation according to the first test parameters corresponding to the first pressure stage. After the first pressure phase ends, the first pressure device is paused, and the second pressure device is invoked to perform data access operations according to the second test parameters corresponding to the second pressure phase. After the second pressure phase ends, the invocation of the second pressure device is paused, and the step of invoking the first pressure device to perform data access operations according to the first test parameters corresponding to the first pressure phase continues until the storage device under test meets the test termination requirements.

11. An electronic device, wherein, include: The device includes a memory, a processor, and a communication interface; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor performs the performance testing method as described in any one of claims 1 to 9 or claim 10.

12. A non-transitory machine-readable storage medium, wherein, The non-transitory machine-readable storage medium stores executable code that, when executed by a processor of an electronic device, causes the processor to perform the performance testing method as described in any one of claims 1 to 9 or claim 10.

13. A computer program product, wherein, include: A computer program, when executed by a processor of an electronic device, causes the processor to perform the performance testing method as described in any one of claims 1 to 9 or claim 10.

14. A performance testing apparatus, wherein, include: The filling module is used to fill the storage device under test with data; An execution module is configured to perform a data access operation on the storage device under test according to a first test parameter corresponding to a first pressure stage when the storage device under test meets a first set filling condition, and to perform a data access operation on the storage device under test according to a second test parameter corresponding to a second pressure stage when the data access operation corresponding to the first pressure stage meets a first switching condition, wherein the first test parameter and the second test parameter are different. The detection module is used to detect that the data access operation corresponding to the second pressure stage meets the second switching condition, and continue to execute the step of performing the data access operation according to the first test parameters corresponding to the first pressure stage until the storage device under test meets the test termination requirements.

15. The apparatus according to claim 14, wherein, The execution module is specifically used for: If the storage device under test meets the first set filling condition, a target write request is sent to the storage device under test according to the first test parameters; The device further includes a sending module, configured to continue sending a target write request corresponding to the first test parameter to the storage device under test when the data access operation corresponding to the first pressure stage is detected to not meet the first switching condition.

16. The apparatus according to claim 15, wherein, The sending module is also used for: After any target write request response is completed, at least one performance metric of the storage device under test is collected. The first test parameter is adjusted based on the at least one performance metric data, so that the target write request can continue to be sent to the storage device under test using the adjusted first test parameter.

17. The apparatus according to claim 15, wherein, The first test parameter includes a first rate limiting quantity; the sending module is specifically used for: If the storage device under test meets the first set filling condition, a target write request is sent to the storage device under test according to the first current limit quantity; If the data access operation corresponding to the first pressure phase is not satisfied with the first switching condition, count whether the number of sent requests has reached the first rate limit. If the number of requests sent has not reached the first rate limit, continue to send target write requests corresponding to the first test parameters to the storage device under test.

18. The apparatus according to claim 17, wherein, The sending module is also used for: If the number of sent requests reaches the first rate limit, the sending of target write requests to the storage device under test will be suspended. After any first request response is completed, update the number of requests sent; If it is detected that the number of requests sent has not reached the first rate limit, then continue to send target write requests corresponding to the first test parameters to the storage device under test.

19. The apparatus according to any one of claims 14-18, wherein, The filling module is specifically used for: Based on multiple checkpoints generated during model training, a target task is generated; the target task includes the first set filling condition, the first test parameter corresponding to the first stress stage, and the second test parameter corresponding to the second stress stage. The target task is to be performed to populate the storage device under test with data according to the target water level.

20. The apparatus according to any one of claims 14-18, wherein, The first test parameters include the first request write duration, the first bandwidth, and the first rate limit; the second test parameters include the second request write duration, the second bandwidth, and the second rate limit. The execution module is specifically used for: If the first request write duration is detected to be greater than the first duration threshold, then it is determined that the data access operation corresponding to the first stress phase meets the first switching condition. Data access operations are then performed on the storage device under test according to the second request write duration, the second bandwidth, and the second rate limiting quantity corresponding to the second stress phase. The step of detecting that the data access operation corresponding to the second stress phase meets the second switching condition includes: if the second request write duration is detected to be greater than the second duration threshold, then it is determined that the data access operation corresponding to the second stress phase meets the second switching condition.