Introduction to several tools for stress testing on Linux disks or other types of servers

(I) DD: reads data from one source and writes it to another destination address in bits.

[root@linwaterbin ~]# dd if=/dev/zero of=/home/oracle/disktest bs=1M count=512 conv=fdatasync

Note
If: from where
Of: where to go
BS: size of data blocks read each time
Count: Number of BS read
Conv: eliminates the impact of Linux memory cache and ensures that data is directly written to the disk.

Test procedure:
We recommend that you run the command multiple times to get the average value and clear the Linux cache before each run.

[root@linwaterbin ~]# echo 3 > /proc/sys/vm/drop_caches[root@linwaterbin ~]# dd if=/dev/zero of=/home/oracle/disktest bs=1M count=512 conv=fdatasync512+0 records in512+0 records out536870912 bytes (537 MB) copied, 27.4893 seconds, 19.5 MB/s[root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# echo 3 > /proc/sys/vm/drop_caches[root@linwaterbin ~]# dd if=/dev/zero of=/home/oracle/disktest bs=1M count=512 conv=fdatasync512+0 records in512+0 records out536870912 bytes (537 MB) copied, 17.3697 seconds, 30.9 MB/s[root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# echo 3 > /proc/sys/vm/drop_caches[root@linwaterbin ~]# dd if=/dev/zero of=/home/oracle/disktest bs=1M count=512 conv=fdatasync512+0 records in512+0 records out536870912 bytes (537 MB) copied, 14.9991 seconds, 35.8 MB/s[root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# [root@linwaterbin ~]# echo 3 > /proc/sys/vm/drop_caches[root@linwaterbin ~]# dd if=/dev/zero of=/home/oracle/disktest bs=1M count=512 conv=fdatasync512+0 records in512+0 records out536870912 bytes (537 MB) copied, 15.2154 seconds, 35.3 MB/s

19.5 Mb/s + 30.9 Mb/s + 35.8 Mb/s + 35.3 Mb/s add these values to the average

(Ii) hdparm
There are two parameters:
-T: disk Performance Detection
-T: memory cache Performance Detection
We also recommend that you average the number of tests.

[root@linwaterbin ~]# hdparm -t /dev/sda/dev/sda: Timing buffered disk reads:  226 MB in  3.02 seconds =  74.82 MB/sec[root@linwaterbin ~]# hdparm -t /dev/sda/dev/sda: Timing buffered disk reads:  244 MB in  3.10 seconds =  78.59 MB/sec[root@linwaterbin ~]# hdparm -t /dev/sda/dev/sda: Timing buffered disk reads:  246 MB in  3.12 seconds =  78.87 MB/sec

The above two tools only return read/write performance, and the test is relatively simple.

Use Bonnie ++ or Iozone to generate detailed disk reports
Configure the following before installation:

Download the corresponding version from the repoforge website

[root@linwaterbin Desktop]# rpm -ivh --nodeps rpmforge-release-0.5.2-2.el5.rf.i386.rpm warning: rpmforge-release-0.5.2-2.el5.rf.i386.rpm: Header V3 DSA signature: NOKEY, key ID 6b8d79e6Preparing...                ########################################### [100%]   1:rpmforge-release       ########################################### [100%][root@linwaterbin Desktop]# cd /etc/yum.repos.d[root@linwaterbin yum.repos.d]# lsbase.repo         mirrors-rpmforge-extras   redhat.repo          rpmforge.repomirrors-rpmforge  mirrors-rpmforge-testing  rhel-debuginfo.repo

(3) Bonnie ++
Installation:

[root@linwaterbin yum.repos.d]# yum install -y bonnie++

First, describe the common parameters in the command.
-D path for generating the test file
-S: size of the test file generated, in MB (if the-R parameter is not used, the file size must be at least twice the physical memory of the system)
-M machine name, which can be considered as the solution name for this test. The default is the host name of the local machine.
-R memory size: Specifies the memory size. In this way, you can use the-S parameter to create an R * 2 file, which is usually used to shorten the test time.
However, it should be noted that the test results may be inaccurate due to the memory cache.
-X number of tests
-U indicates the owner and group of the test file. The default value is the current user and the current group that executes Bonnie ++.
-G indicates the group of test files. The default value is the current group that executes Bonnie ++.
-B calls the fsync () function every time a file is written. It is suitable for testing the mail server or database server, which usually requires synchronization operations,
Without this parameter, it is more suitable for testing the efficiency of copying files or compiling operations.

[root@linwaterbin ~]# bonnie++  -s 512 -r 256 -u root

Main output:

Version  1.96       ------Sequential Output------ --Sequential Input- --Random-Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CPlinwaterbin    512M   217  99 32403  14 13798   7   378  99 49235  10 347.0   3Latency               154ms    2218ms    2099ms     125ms   63304us    2672msVersion  1.96       ------Sequential Create------ --------Random Create--------linwaterbin         -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP                 16 15508  52 +++++ +++ +++++ +++ 27215  91 +++++ +++ +++++ +++Latency             43819us   20118us   19580us   19834us   19699us   20435us

(1) sequential output: write operation
① Per CHR: character
② Block: Block
(2) sequential input: read operation
(3) K/sec: Speed (K/sec)
(4) % CP: CPU usage


(Iv) Iozone
Supports concurrent multi-process testing
Ability to output to excel for drawing
It is also the most widely used tool for routine stress testing.
Installation:

[root@linwaterbin ~]# yum install -y iozone

Parameter description:
-L: Minimum number of processes
-U: Maximum number of processes
-R: Basic read/write unit, subject to the block size of the test object
For example, if the Oracle block is set to 8 K, you can set it to 8 K.
-S: It is consistent with the meaning of the S parameter of Bonnie ++.
If this value is too small, the test results will be affected.
Because many programs have run out of memory
-F: cached files

[root@linwaterbin ~]# iozone -l 1 -u 1 -r 8K -s 128M        Record Size 8 KB        File size set to 131072 KB        Command line used: iozone -l 1 -u 1 -r 8K -s 128M        Output is in Kbytes/sec        Time Resolution = 0.000001 seconds.        Processor cache size set to 1024 Kbytes.        Processor cache line size set to 32 bytes.        File stride size set to 17 * record size.        Min process = 1         Max process = 1         Throughput test with 1 process        Each process writes a 131072 Kbyte file in 8 Kbyte records        Children see throughput for  1 initial writers  =  125057.24 KB/sec        Parent sees throughput for  1 initial writers   =   30640.70 KB/sec        Min throughput per process                      =  125057.24 KB/sec         Max throughput per process                      =  125057.24 KB/sec        Avg throughput per process                      =  125057.24 KB/sec        Min xfer                                        =  131072.00 KB        Children see throughput for  1 rewriters        =  513780.34 KB/sec        Parent sees throughput for  1 rewriters         =   31989.50 KB/sec        Min throughput per process                      =  513780.34 KB/sec         Max throughput per process                      =  513780.34 KB/sec        Avg throughput per process                      =  513780.34 KB/sec        Min xfer                                        =  131072.00 KB        Children see throughput for  1 readers          =  889758.12 KB/sec        Parent sees throughput for  1 readers           =  849615.75 KB/sec        Min throughput per process                      =  889758.12 KB/sec         Max throughput per process                      =  889758.12 KB/sec        Avg throughput per process                      =  889758.12 KB/sec        Min xfer                                        =  131072.00 KB

The value here is very large because our S parameter settings are a little small.