A storage test method based on cloud rendering

Abstract

The present invention discloses a storage testing method based on cloud rendering, comprising the following steps: creating a K-frame rendering task in a 3T rendering scene, and submitting the K-frame rendering task to a scheduler through a cloud rendering platform; after the scheduler assigns the K-frame rendering task to K server nodes respectively, the K server nodes concurrently execute the frame rendering task; while the frame rendering task is in progress, continuously performing read and write operations on the storage under test; the storage under test records performance change information of the storage under test when the frame rendering task is performed; after the K-frame rendering task is completed, optimizing and adjusting the configuration information of the storage under test according to the performance change information. The present invention can optimize the configuration information of the storage under test based on the test results so that it can meet the rendering requirements of the K-frame rendering task, improve the performance of the storage under test, reduce online crashes, and thus optimize rendering efficiency.

Description

Technical Field

[0001] The present invention relates to the field of cloud rendering, and in particular to a storage testing method based on cloud rendering. Background Art

[0002] The rendering efficiency of a cloud render farm is generally affected by the storage performance of the cloud render farm. Therefore, it is necessary to test the storage performance of the cloud render farm to ensure the rendering efficiency of the cloud render farm.

[0003] Existing technologies typically use tools like fio to test storage performance, but these tools are not compatible with cloud rendering platforms, making it impossible to test storage performance on cloud rendering platforms. Consequently, there's no guarantee that storage performance will support cloud rendering platforms, which can easily lead to online crashes and reduced rendering efficiency.

[0004] Therefore, the prior art has defects and needs to be improved. Summary of the Invention

[0005] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a storage testing method based on cloud rendering to solve the problem in the existing technology that storage performance testing of cloud rendering storage cannot be realized, which easily leads to online crashes and reduced rendering efficiency during the cloud rendering process.

[0006] The technical solution of the present invention is as follows: A storage testing method based on cloud rendering comprises the following steps:

[0007] S1: Place a 3T rendering scene on the storage under test, and create a K-frame rendering task in the 3T rendering scene.

[0008] S2: Submitting the K-frame rendering task in the 3T rendering scene to the scheduler through the cloud rendering platform.

[0009] S3: The scheduler assigns the K frame rendering tasks to K server nodes respectively, and the K server nodes execute the frame rendering tasks concurrently.

[0010] S4: When performing corresponding frame rendering tasks, the K server nodes simultaneously and continuously perform read and write operations of the frame rendering tasks on the tested storage.

[0011] S5: The tested storage records performance change information of the tested storage when the K server nodes perform read and write operations on the tested storage during the frame rendering tasks.

[0012] S6: After the K-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage according to the performance change information, and repeats steps S1 to S5 after the optimization and adjustment are completed.

[0013] Furthermore, the step S5 further includes: after the K frame rendering tasks are completed, comparing the rendering duration of each frame rendering task in the K frame rendering tasks with its corresponding standard duration, and outputting the duration comparison result.

[0014] Furthermore, the step S6 is: after the K-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage according to the performance change information and the duration comparison result, and repeats steps S1 to S5 after the optimization and adjustment are completed.

[0015] Furthermore, the performance change information includes: total rendering time, failure rate of single-frame rendering tasks, CPU parameter values, IO values, and traffic values stored under test.

[0016] Furthermore, the configuration information of the tested storage includes: disk type, data structure, storage and backup methods of the tested storage.

[0017] Furthermore, the scheduler is a munu scheduler.

[0018] Using the above solution, the present invention provides a storage testing method based on cloud rendering, which has the following beneficial effects:

[0019] 1. It can quickly test the storage performance of the storage under test for cloud rendering, and prevent the storage under test from crashing due to complex scenes during frame rendering tasks;

[0020] 2. Based on the test results, the configuration information of the tested storage can be optimized to meet the rendering requirements of the K-frame rendering task, improve the performance of the tested storage, reduce online crashes, and thus optimize rendering efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a test flow chart of the tested storage of the present invention. DETAILED DESCRIPTION

[0022] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] Please refer to Figure 1 The present invention provides a storage testing method based on cloud rendering, comprising the following steps:

[0024] S1: Place a 3TB rendering scene on the storage device under test and create K frame rendering tasks within the 3TB rendering scene. Specifically, in this embodiment, K = 3000. The 3TB rendering scene does not contain large files, small files, or special files with different file suffixes, ensuring sufficient read and write capacity for the storage device under test.

[0025] It should be noted that, generally, when conducting performance tests on the storage under test, the number of frame rendering tasks established in the 3T rendering scene is in an increasing state. For example, 800 rendering frame tasks are first created to perform the operation process of steps S1 to S6. After meeting the requirements, 1200, ..., 3000 frame rendering task tests are created in sequence, and optimization and testing are carried out step by step.

[0026] S2: Submitting the 3000-frame rendering task in the 3T rendering scene to the scheduler through the cloud rendering platform. That is, submitting the 3000-frame rendering task to the cloud rendering platform.

[0027] S3: The scheduler distributes the 3000 frame rendering tasks to 3000 server nodes respectively, and the 3000 server nodes execute the frame rendering tasks concurrently. Specifically, the scheduler is munu scheduler.

[0028] S4: While performing the corresponding frame rendering tasks, the 3,000 server nodes simultaneously and continuously read and write the frame rendering tasks to the storage under test. During this process, each server node will first call up the rendering software on the machine to prepare for the subsequent rendering operations; specifically, the rendering software can be CG rendering software.

[0029] S5: The tested storage records the performance change information of the tested storage when the 3000 server nodes perform read and write operations on the tested storage during the frame rendering tasks. At this time, the 3000 server nodes perform read and write operations on the 3T rendering scene at the same time, that is, read and write operations on the tested storage for frame rendering tasks; during the frame rendering task, the performance change information of the tested storage is constantly changing, so the tested storage will record the changes in storage read and write performance and generate performance change information. Specifically, the performance change information includes: the total rendering time, the failure rate of a single frame rendering task, the CPU parameter value, IO value, and traffic value of the tested storage, and the total rendering time is generated after the 3000 frame rendering tasks are completed.

[0030] S6: After the 3000-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage according to the performance change information, and repeats steps S1 to S5 after the optimization and adjustment is completed. The configuration information of the tested storage includes: the disk type, data structure, storage and backup methods of the tested storage.

[0031] The present invention provides a storage testing method based on cloud rendering. By placing a 3T rendering scene on the tested storage, which includes large files, small files, and various special files, the tested storage can cover various special rendering scene type data, which can prevent the tested storage from crashing due to complex scenes during the frame rendering task. Each frame rendering task corresponds to a server node, and 3000 server nodes start rendering at the same time and perform read and write operations on the tested storage at the same time. At this time, the tested storage needs to exchange information with 3000 server nodes at the same time, so its performance requirements are very high. The requirements are higher. During this process, its performance changes will be monitored in real time and its performance change information will be recorded. After the 3000-frame rendering task is completed, the information recorded in its performance change information can be used to check whether the problems existing in the tested storage when performing the 3000-frame rendering task at the same time meet the requirements. If the requirements are not met, the configuration information of the tested storage will be optimized according to the performance change information to enable it to meet the simultaneous rendering requirements of the 3000-frame rendering task, improve the performance of the tested storage, ensure that the storage performance of the tested storage can support the cloud rendering platform, reduce online crashes, and thus optimize rendering efficiency.

[0032] Specifically, in this embodiment, step S5 further includes: after the K-frame rendering task is completed, comparing the rendering duration of each frame rendering task in the K-frame rendering task with its corresponding standard duration, and outputting the duration comparison result. It should be noted that the standard duration is the optimal rendering duration corresponding to the 3000-frame rendering task, which is generally data obtained through preliminary testing.

[0033] Step S6 is as follows: after the K-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage based on the performance change information and the duration comparison result, and repeats steps S1 to S5 after the optimization and adjustment are completed. After the duration comparison result is output, the duration comparison result and the performance change information can be comprehensively analyzed to optimize the configuration information of the tested storage. At the same time, after optimizing some basic parameters of the tested storage based on the performance change information, other performance-related configuration information of the tested storage can be further optimized based on the duration comparison result.

[0034] In summary, the present invention provides a storage testing method based on cloud rendering, which has the following beneficial effects:

[0035] 1. It can quickly test the storage performance of the storage under test for cloud rendering, and prevent the storage under test from crashing due to complex scenes during frame rendering tasks;

[0036] 2. Based on the test results, the configuration information of the tested storage can be optimized to meet the rendering requirements of the K-frame rendering task, improve the performance of the tested storage, reduce online crashes, and thus optimize rendering efficiency.

[0037] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A storage testing method based on cloud rendering, characterized in that: The following steps are involved: S1: Place a 3T rendering scene on the storage under test and create a K-frame rendering task in the 3T rendering scene; S2: Submitting the K-frame rendering task in the 3T rendering scene to the scheduler through the cloud rendering platform; S3: The scheduler assigns the K frame rendering tasks to K server nodes respectively, and the K server nodes execute the frame rendering tasks concurrently; S4: When performing corresponding frame rendering tasks, the K server nodes simultaneously and continuously perform read and write operations of the frame rendering tasks on the tested storage; S5: The tested storage records performance change information of the tested storage when the K server nodes perform read and write operations on the tested storage during the frame rendering task; S6: After the K-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage according to the performance change information, and repeats steps S1 to S5 after the optimization and adjustment are completed; The performance change information includes: total rendering time, failure rate of single-frame rendering tasks, CPU parameter values, IO values, and traffic values of the tested storage; The scheduler is a munu scheduler.

2. A storage testing method based on cloud rendering according to claim 1, characterized in that: The step S5 further includes: after the K frame rendering tasks are completed, comparing the rendering duration of each frame rendering task in the K frame rendering tasks with its corresponding standard duration, and outputting the duration comparison result.

3. The storage testing method based on cloud rendering according to claim 2, characterized in that: The step S6 is: after the K-frame rendering task is completed, the tested storage optimizes and adjusts the configuration information of the tested storage according to the performance change information and the duration comparison result, and repeats steps S1 to S5 after the optimization and adjustment are completed.

4. The storage testing method based on cloud rendering according to claim 1, characterized in that: The configuration information of the tested storage includes: the disk type, data structure, storage and backup mode of the tested storage.

Images