WO2022078009A1

WO2022078009A1 - Apparatus and method for testing performance of computer storage system, and storage medium thereof

Info

Publication number: WO2022078009A1
Application number: PCT/CN2021/109448
Authority: WO
Inventors: 于泉泉; 王鹏; 王焕超; 刘闻禹; 孙珑玲; 闫玉婕
Original assignee: 苏州浪潮智能科技有限公司
Priority date: 2020-10-13
Filing date: 2021-07-30
Publication date: 2022-04-21
Also published as: CN112269696A

Abstract

An apparatus and a method for testing the performance of a computer storage system, and a storage medium thereof, relating to the technical field of computer storage systems, the method for testing the performance of a computer storage system comprising: dividing bandwidth topology nodes on the basis of the links between the devices in a storage topology and dividing IOPS topology nodes on the basis of the devices in the storage topology; calculating the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node; on the basis of the highest performance value of the bandwidth topology nodes, performing measurement to obtain a bandwidth bottleneck node; on the basis of the highest performance value of the IOPS topology nodes, performing measurement to obtain an IOPS bottleneck node; setting the highest performance value of the bandwidth bottleneck node or the IOPS bottleneck node as the highest system performance value; and, on the basis of the highest system performance value, acquiring a test result and outputting the test result.

Description

A computer storage system performance testing device, method and storage medium thereof

This application claims the priority of the Chinese patent application filed on October 13, 2020 with the application number 202011090549.6 and the invention titled "A computer storage system performance testing device, method and storage medium", the entire contents of which are Incorporated herein by reference.

technical field

The present invention relates to the technical field of computer storage systems, in particular to a computer storage system performance testing device, method and storage medium thereof.

Background technique

With the rapid development of the Internet and the continuous rise of cloud computing technology, the volume of online business continues to increase. In the face of rapid changes in server technology, how to improve the performance of a storage system has always been an important technical problem. Performance is a key factor in evaluating a storage system. Understanding the performance test method, calculating the rationality of the test results, and mastering the necessary rectification methods are the keys to solving performance problems. There are two main parameters that reflect the performance of the storage system: bandwidth (Throughout) and IOPS (Input/Output Operations Per Second, the number of read and write operations per second). Bandwidth is used to measure the IO (more than 16KB) capability of the storage system to process sequential reads and writes of large data blocks. The unit is MB/s. The higher the bandwidth, the better the performance; IOPS is used to measure the IO of the storage system to process random read and write small data blocks. (Below 16KB) capability, that is, the number of read and write (I/O) operations per second. The higher the IOPS, the stronger the storage system's ability to handle IO. At the same time, the performance test is a test for a specific device or a specific topology. When testing the performance of a specific component, ensure that the performance of other equipment in the system is better than that of the test equipment, and ensure that the test equipment is the only bottleneck, so as to test the maximum performance of the equipment; when testing the performance of the system, ensure that all interfaces and components work In the best state, in order to measure the limit performance value of the entire system and find the bottleneck point that limits the performance, in other words, the performance test is not a single component test, but a comprehensive capability test of the entire storage system. Due to this limitation, the methods for performance testing vary widely, but there are few theoretical calculation methods for performance under storage topology.

The traditional storage system performance test builds a simple empirical database or mathematical model scheme, relies on the experience data of testers, or the experience accumulation of the existing test database, and does not analyze the performance parameters and link relationships of the devices under the topology, resulting in:

1. Poor flexibility. The current model is only valid for a specific topology or several types of topology. If it is replaced with other topology, the mathematical model needs to be rebuilt according to the actual measurement results.

2. The accuracy is not high. When the device under the storage system is changed, for example, the hard disk is changed from HDD (Hard Disk Drive, hard disk drive) to SSD (Solid State Disk, solid state disk), even if the topology is the same, the result changes greatly.

3. The verification cost is high. To verify the correctness of the data, it is necessary to cover the performance tests under multiple configurations and multiple topologies, and to provide sufficient measured data.

Therefore, this paper proposes a computer storage system performance testing device, method and storage medium.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a computer storage system performance testing device, method and storage medium, so as to alleviate the technical problems of poor flexibility, low accuracy and high verification cost in the prior art.

In a first aspect, a method for testing the performance of a computer storage system provided by the present invention includes:

The bandwidth topology nodes are divided according to the links between each device in the storage topology, and the IOPS topology nodes are divided according to each device in the storage topology;

Obtain the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node;

Calculate based on the highest performance value of each bandwidth topology node to obtain the bandwidth bottleneck node;

Calculate based on the highest performance value of each IOPS topology node to obtain the IOPS bottleneck node;

Take the highest performance value of the bandwidth bottleneck node or IOPS bottleneck node as the highest system performance value;

A test result is obtained based on the highest value of the system performance, and the test result is output.

Further, the steps of calculating and obtaining bandwidth bottleneck nodes based on the highest performance value of each bandwidth topology node include:

Calculate the bandwidth bottleneck node by the formula BW _Bottleneck =Minimum(BW _Node1 ,BW _Node2 ,...,BW _NodeN );

Among them, Bottleneck is the bottleneck node, BW _Bottleneck is the highest performance value of the bandwidth bottleneck node, and BW _Nodei is the highest bandwidth value of each bandwidth topology node.

Further, the steps of calculating and obtaining the IOPS bottleneck node based on the highest performance value of each IOPS topology node include:

Calculate the IOPS bottleneck node through the formula IOPS _Bottleneck = Minimum (IOPS _Node1 , IOPS _Node2 , ..., IOPS _NodeN );

Among them, Bottleneck is the bottleneck node, IOPS _Bottleneck is the highest performance value of the IOPS bottleneck node, and IOPS _Nodei is the highest IOPS value of each IOPS topology node.

Perform a single-disk IOPS test on each hard disk, and obtain the single-disk bottleneck node as the IOPS bottleneck node.

Further, the steps of performing a single-disk IOPS test on each hard disk to obtain the highest performance value of the single-disk bottleneck node system include:

Calculate the total system delay Latency _Bottleneck by the formula Latency _Bottleneck =Sum(Latency _Node1 , Latency _Node2 ,..., Latency _NodeN ); wherein, Latency _Nodei is the maximum delay value of each IOPS topology node;

through the formula

Calculate the maximum performance value IOPS _Bottleneck of a single-disk bottleneck node; where, Latency _disk is the single-disk latency, and IOPS _disk is the single-disk maximum IOPS value.

Further, the steps of taking the highest performance value of the bandwidth bottleneck node or the IOPS bottleneck node as the highest system performance value include:

Take the lower maximum performance value of the bandwidth bottleneck node and the IOPS bottleneck node as the system performance maximum value.

Further, after the step of obtaining the test result based on the highest value of the system performance and outputting the test result, it also includes:

Compare the test results with the actual test results;

If the comparison difference exceeds the preset value, the configuration parameters of the topology nodes are adjusted, and the steps of calculating the maximum performance value of each bandwidth topology node and the maximum performance value of each IOPS topology node are returned.

In a second aspect, the present invention also provides a computer storage system performance testing device, comprising:

The topology module is used to divide bandwidth topology nodes according to links between various devices in the storage topology, and divide IOPS topology nodes according to each device in the storage topology;

The calculation module is used to obtain the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node;

The calculation module is used to calculate and obtain the bandwidth bottleneck node based on the highest performance value of each bandwidth topology node, and obtain the IOPS bottleneck node based on the calculation based on the highest performance value of each IOPS topology node, and use the bandwidth bottleneck node or the highest performance value of the IOPS bottleneck node as the The highest value of system performance;

An output module, configured to obtain a test result based on the highest value of the system performance, and output the test result.

A computer-readable storage medium provided by the present invention stores machine-executable instructions. When the computer-executable instructions are invoked and executed by a processor, the computer-executable instructions cause the The processor runs the computer storage system performance testing method.

A computer storage system performance testing device, method, and storage medium provided by the embodiments of the present invention divide bandwidth topology nodes according to links between various devices in the storage topology, divide IOPS topology nodes according to each device in the storage topology, and Obtain the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node, obtain the bandwidth bottleneck node and IOPS bottleneck node, use the highest performance value of the bandwidth bottleneck node or IOPS bottleneck node as the highest performance value of the system, and finally output Test Results. Therefore, the present invention can effectively obtain test results for various topology models, and solve the problem of poor flexibility in the prior art. When the device under the storage system is changed, the test accuracy can also be guaranteed, and it does not need to cover multiple configurations and multiple topologies. Saves the verification cost of the test.

Description of drawings

In order to illustrate the specific embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the specific embodiments or the prior art. Obviously, the accompanying drawings in the following description The drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

1 is a flowchart of a storage system performance testing method provided by an embodiment of the present invention;

2 is a schematic diagram of a performance method under a storage topology provided by an embodiment of the present invention;

3 is a bandwidth node diagram of a storage topology provided by an embodiment of the present invention;

FIG. 4 is an IOPS node diagram of a storage topology provided by an embodiment of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of them. example. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The terms "comprising" and "having" mentioned in the embodiments of the present invention and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes other unlisted steps or units, or optionally also Include other steps or units inherent to these processes, methods, products or devices.

An embodiment of the present invention provides a technical solution, a method for testing the performance of a computer storage system, including:

Calculate the bandwidth bottleneck node by the formula BW _Bottleneck =Minimum(BW _Node1 ,BW _Node2 ,...,BW _NodeN ) (——Formula 1);

Calculate the IOPS bottleneck node by the formula IOPS _Bottleneck =Minimum(IOPS _Node1 , IOPS _Node2 , ..., IOPS _NodeN ) (——Formula 2);

Calculate the total delay Latency _Bottleneck of the system by calculating the formula Latency _Bottleneck =Sum(Latency _Node1 ,Latency _Node2 ,...,Latency _NodeN ) (——Formula 3); wherein, Latency _Nodei is the maximum delay value of each IOPS topology node;

through the formula

The test results are compared with the actual test results; the running performance test refers to the process of using performance test tools (such as IOMeter, FIO, etc.) and outputting the performance test results, and comparing the theoretical calculation results and the actual measurement results. A theoretical model for tuning performance computing.

If the comparison difference exceeds the preset value, the performance of the topology node is adjusted, and the steps of calculating the maximum performance value of each bandwidth topology node and the maximum performance value of each IOPS topology node are returned. For example, in the previous test, the operating parameters of the hard disk were set in a way suitable for sequential reading and writing. In this step, the operating parameters of the hard disk can be adjusted to a configuration more suitable for random reading and writing, so that the hard disk can be used in subsequent tests. , can play a higher performance. For another example, the parameters of the BIOS (Basic Input Output System) of the motherboard can be adjusted, so that the motherboard can exert higher performance in subsequent tests. According to multiple performance tests, the method of calculating the performance under the storage topology by the node method is basically consistent with the measured results, which proves that the theoretical analysis method is effective.

In the method for testing the performance of a computer storage system provided by the embodiment of the present invention, bandwidth topology nodes are divided according to links between various devices in the storage topology, IOPS topology nodes are divided according to each device in the storage topology, and the performance of each bandwidth topology node is obtained. The highest value and the highest performance value of each IOPS topology node are obtained to obtain the bandwidth bottleneck node and IOPS bottleneck node. Therefore, the present invention can effectively obtain test results for various topology models, and solve the problem of poor flexibility in the prior art. When the device under the storage system is changed, the test accuracy can also be guaranteed, and it does not need to cover multiple configurations and multiple topologies. Saves the verification cost of the test.

An embodiment of the present invention provides a computer storage system performance testing device, including:

Embodiments of the present invention provide a computer-readable storage medium, where the computer-readable storage medium stores machine-executable instructions, and when the computer-executable instructions are invoked and executed by a processor, the computer-executable instructions cause all The processor runs the computer storage system performance testing method.

It should be noted in this embodiment that, as shown in FIG. 1 , nodes are divided according to the storage topology. The topology of the storage system is formed at the beginning of the system design. At this time, the storage topology of the system can be directly divided into nodes. For example, the typical storage topology node division is shown in the following figure: The motherboard CPU (central processing unit, central processing unit) is connected to the SAS Card (Serial Attached SCSI) through the PCIe (peripheral component interconnect express, a high-speed serial computer expansion bus standard) slot. (Small Computer System Interface, small computer system interface) Card, serial small computer system interface card), the SAS card is connected to the Expander (expander) cascade backplane through a SAS cable, and the Expander cascade backplane is connected to 12 SATA SSD hard drives are connected via connectors:

The theoretical bandwidth node topology is shown in the following figure: Node1 is the PCIe link between the motherboard PCIe and the SAS Card; Node2 is the SAS link between the SAS card and the Expander cascading backplane; Node3 is the Expander cascading backplane and 12 cards All SATA (Serial Advanced Technology Attachment) links between hard drives.

As shown in Figure 2, each device unit of the storage system is divided into multiple different nodes (NODEs) according to their positions in the topology; the arrows between the nodes represent the flow of storage IO; when calculating the theoretical maximum bandwidth, the nodes represent the storage system The exchange rate of high-speed signals between upstream and downstream devices in the topology; when calculating the theoretical maximum IOPS, a node represents a device unit in the storage system topology;

Among them, the highest performance value of each node is an inherent attribute of each node equipment unit, such as factory parameters, design parameters, etc. In this step, the factory batch, model, and design parameters of the node equipment unit can be directly obtained.

Each node needs to calculate the theoretical maximum bandwidth (BW _NodeN ) and maximum IOPS (IOPS _NodeN ).

When the theoretical maximum bandwidth BW _NodeN of a node is equal to BW _Bottleneck , it means that the bandwidth performance bottleneck of the current storage system is node N, and the maximum bandwidth of node N is the maximum bandwidth value of the storage system; when the theoretical maximum IOPS value of a node, IOPS _NodeN is equal to IOPS _Bottleneck , it indicates that the IOPS performance bottleneck of the current storage system is node N, and the maximum IOPS of node N is the maximum IOPS value of the storage system.

In the storage system topology, since IOPS=IOdepth/Latency, that is, the system latency (Latency) is inversely proportional to the IOPS when the IOdepth remains unchanged, so the system IOPS can also be characterized by the system latency.

Latency _Bottleneck =Sum(Latency _Node1 ,Latency _Node2 ,...,Latency _NodeN )——Equation 3

As shown in Figure 3-4, calculate the maximum bandwidth and maximum IOPS of each node.

Get the performance bottleneck node and the theoretical maximum bandwidth and maximum IOPS value.

According to formula 1, the bandwidth bottleneck node is Node1 (ie, the PCIe link between the motherboard PCIe slot and the SAS card), and the maximum bandwidth is 6400MB/s;

According to formula 2, the IOPS bottleneck node is Node4 (that is, the total 4K performance of 12 hard disks), and the maximum IOPS is 1164 K IOPS.

It can be seen that the performance bandwidth and IOPS may correspond to different bottleneck nodes respectively. This is not contradictory, because the performance test items mainly include two types of bandwidth and IOPS, and the two test methods are different.

Running performance testing refers to the process of using performance testing tools (such as IOMeter, FIO, etc.) and outputting performance testing results, comparing the theoretical calculation results with the actual measurement results, and then correcting and tuning the theoretical model of performance calculation.

According to multiple performance tests, the method of calculating the performance under the storage topology by the node method is basically consistent with the measured results, which proves that the theoretical analysis method is effective.

For example, a set of parallel performance bandwidth test data comparisons under this topology:

For example, a set of parallel performance IOPS test data comparisons under this topology:

Different from the results of the parallel test, when the single-disk performance test IOPS is measured for all hard disks one by one, in theory, the hard disks in this topology should become the IOPS bottleneck point in this topology. However, the actual measurement results show that the hard disk cannot reach the maximum IOPS value specified by the manufacturer during the 4K RR performance test. When testing the IOPS of a single disk in this topology, Equation 3 needs to be used to solve it.

According to formula 3, in the single-disk performance test, the theoretical latency bottleneck of the system is often the SSD or HDD hard disk; the total latency is: Latency _bottleneck =Sum(5,60,50,400)=515us. The ratio of the measured IOPS value to the IOPS of a single disk can be calculated by the ratio of the theoretical delay of the hard disk to the total delay:

i.e. 400/515=77.7%

The theoretical IOPS node topology is shown in the following figure: Node1 is the main board, Node2 is the SAS Card, Node3 is the Expander cascaded backplane, and Node4 is all 12 SSDs.

It should be noted that like numerals and letters refer to like items in the following figures, so once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. The indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of the invention is usually placed in use, only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying The device or element referred to must have a particular orientation, be constructed and operate in a particular orientation, and therefore should not be construed as limiting the invention. Furthermore, the terms "first", "second", "third", etc. are only used to differentiate the description and should not be construed as indicating or implying relative importance.

In the description of the present invention, it should also be noted that, unless otherwise expressly specified and limited, the terms "arranged", "installed", "connected" and "connected" should be understood in a broad sense, for example, it may be a fixed connection, It can also be a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or the internal communication between the two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood in specific situations.

Claims

A method for testing computer storage system performance, comprising:

The bandwidth topology nodes are divided according to the links between each device in the storage topology, and the IOPS topology nodes are divided according to each device in the storage topology;

Obtain the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node;

Calculate based on the highest performance value of each bandwidth topology node to obtain the bandwidth bottleneck node;

Calculate based on the highest performance value of each IOPS topology node to obtain the IOPS bottleneck node;

Take the highest performance value of the bandwidth bottleneck node or IOPS bottleneck node as the highest system performance value;

A test result is obtained based on the highest value of the system performance, and the test result is output.
The method for testing computer storage system performance according to claim 1, wherein the step of measuring and obtaining a bandwidth bottleneck node based on the highest performance value of each bandwidth topology node, comprises:

Calculate the bandwidth bottleneck node by the formula BW Bottleneck =Minimum(BW Node1 ,BW Node2 ,...,BW NodeN );

Among them, Bottleneck is the bottleneck node, BW Bottleneck is the highest performance value of the bandwidth bottleneck node, and BW Nodei is the highest bandwidth value of each bandwidth topology node.
The method for testing computer storage system performance according to claim 1, wherein the step of calculating and obtaining an IOPS bottleneck node based on the highest performance value of each IOPS topology node, comprises:

Calculate the IOPS bottleneck node through the formula IOPS Bottleneck = Minimum (IOPS Node1 , IOPS Node2 , ..., IOPS NodeN );

Among them, Bottleneck is the bottleneck node, IOPS Bottleneck is the highest performance value of the IOPS bottleneck node, and IOPS Nodei is the highest IOPS value of each IOPS topology node.
The method for testing computer storage system performance according to claim 1, wherein the step of measuring and obtaining an IOPS bottleneck node based on the highest performance value of each IOPS topology node, comprises:

Perform a single-disk IOPS test on each hard disk, and obtain the single-disk bottleneck node as the IOPS bottleneck node.
The method for testing computer storage system performance according to claim 4, wherein the step of performing a single-disk IOPS test on each hard disk to obtain the highest performance value of the single-disk bottleneck node system comprises:

Calculate the total system delay Latency Bottleneck by the formula Latency Bottleneck =Sum(Latency Node1 , Latency Node2 ,..., Latency NodeN ); wherein, Latency Nodei is the maximum delay value of each IOPS topology node;

through the formula
Calculate the maximum performance value IOPS Bottleneck of a single-disk bottleneck node; where, Latency disk is the single-disk latency, and IOPS disk is the single-disk maximum IOPS value.
The method for testing computer storage system performance according to claim 1, wherein the step of using the performance maximum value of the bandwidth bottleneck node or the IOPS bottleneck node as the system performance maximum value comprises:

Take the lower maximum performance value of the bandwidth bottleneck node and the IOPS bottleneck node as the system performance maximum value.
The method for testing computer storage system performance according to claim 1, wherein after the step of obtaining a test result based on the highest value of the system performance and outputting the test result, the method further comprises:

Compare the test results with the actual test results;

If the comparison difference exceeds the preset value, the configuration parameters of the topology nodes are adjusted, and the steps of calculating the maximum performance value of each bandwidth topology node and the maximum performance value of each IOPS topology node are returned.
A computer storage system performance testing device, characterized in that it includes:

The topology module is used to divide bandwidth topology nodes according to links between various devices in the storage topology, and divide IOPS topology nodes according to each device in the storage topology;

The calculation module is used to obtain the highest performance value of each bandwidth topology node and the highest performance value of each IOPS topology node;

The calculation module is used to calculate and obtain the bandwidth bottleneck node based on the highest performance value of each bandwidth topology node, and to obtain the IOPS bottleneck node based on the calculation based on the highest performance value of each IOPS topology node, and use the bandwidth bottleneck node or the highest performance value of the IOPS bottleneck node as the The highest value of system performance;

An output module, configured to obtain a test result based on the highest value of the system performance, and output the test result.
A computer-readable storage medium, characterized in that the computer-readable storage medium stores machine-executable instructions that, when invoked and executed by a processor, cause the computer-executable instructions to cause the A processor executes the method of any one of claims 1 to 7.