Sysbench 0.5

Running Scripts

1. Download the script package and save it to the /home directory.

   1 2	cd /home/ wget https://mirrors.huaweicloud.com/kunpeng/archive/kunpeng_solution/database/scripts/sysbench-0.5-script.zip --no-check-certificate

   You can also use a browser to download the script package and upload it to the /home directory. Download URL: https://mirrors.huaweicloud.com/kunpeng/archive/kunpeng_solution/database/scripts/sysbench-0.5-script.zip

2. Decompress the package and modify the script.
   1. Decompress the script package.

      1	unzip sysbench-0.5-script.zip

   2. Go to the script directory.

      1	cd /home/sysbench-0.5-script

   3. Modify the parameters in the runall.sh script as required. See Table 1.

      1	vi runall.sh

      Table 1 Script parameters

      Parameter	Description
      host='192.168.221.156'	Server IP address
      -P 3306	Port number of the database
      user='root'	Username of the database
      password='123456'	Password of the database
      database='sysbench'	Name of the database. Enter a value according to the actual situation.
      count='100'	Number of databases
      size='1000000'	Table size
      numactl -C 80-127	Ensures that the Sysbench command is isolated from the NIC interrupt in CPU.

   4. Modify the parameters of the nine models in sequence. Modify the following parameters as well:
      • Set the number of concurrent threads to 64, 128, 256, 512.
      • Set the execution time of each concurrent task to 120 seconds (TIME=${TIME:-120}).
      • Add --forced-shutdown=1 before run and $CMD.

        1	--forced-shutdown=1

        

      Open the scripts of the nine models.

      1 2 3 4 5 6 7 8 9

      vi test_oltp_distinct.sh vi test_oltp_index.sh vi test_oltp_nonindex.sh vi test_oltp_order.sh vi test_oltp_point.sh vi test_oltp_simple.sh vi test_oltp_sum.sh vi test_oltp_delete.sh vi test_oltp_mix.sh

3. Execute the model script.
   

   If the database is not restarted before the write scenario (INDEX, NONINDEX, or DELETE) is executed, you need to execute the script of the OLTP MIX scenario once for preheating.

   • Method 1: Run the script of each model in sequence. Set CPU core binding and related parameters as required.
     1. Modify the script permission to obtain the execute permission on the file.

        1 2	cd /home/sysbench-0.5-script chmod 777 *

     2. Execute the model script.
        OLTP DISTINCT read scenario:

        1	numactl -C 80-127 ./test_oltp_distinct.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP INDEX write scenario:

        1	numactl -C 80-127 ./test_oltp_index.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP NONINDEX write scenario:

        1	numactl -C 80-127 ./test_oltp_nonindex.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP ORDER read scenario:

        1	numactl -C 80-127 ./test_oltp_order.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP POINT read scenario:

        1	numactl -C 80-127 ./test_oltp_point.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP SIMPLE read scenario:

        1	numactl -C 80-127 ./test_oltp_simple.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP SUM read scenario:

        1	numactl -C 80-127 ./test_oltp_sum.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP DELETE write scenario:

        1	numactl -C 80-127 ./test_oltp_delete.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

        OLTP MIX mixed scenario:

        1	numactl -C 80-127 ./test_oltp_mix.sh -h 192.168.221.156 -P 3306 -u root -p 123456 -D sysbench -C 100 -S 1000000 -t 512 -c run

   • Method 2: Execute the script in full mode.

     1 2 3

     cd /home/sysbench-0.5-script chmod 777 * sh runall.sh > result-arm.txt

   After the model script is executed, the test result is saved in the result-arm.txt file.

Running Commands

Run the commands of the nine models listed in Table 2 respectively. See Table 3 to modify the parameters in the commands as required.