How to Benchmark RadonDB MySQL with TPC‑C on Kubernetes (Step‑by‑Step)

Background: According to the DWorks 2020 report "The Significance of Chinese‑Developed Databases Ranking First in TPC‑C," more than 67.9% of respondents consider TPC‑C results when selecting a database, making performance a critical metric.

Concepts:

Association

The Transaction Processing Performance Council (TPC) is a nonprofit organization that defines benchmark standards, performance and price metrics, and publishes test results.

Standard

TPC‑C is the OLTP benchmark that evaluates online transaction processing systems and provides reference data for solution selection.

Tool

TPCC‑MySQL is a Percona‑derived implementation of the TPC‑C specification for MySQL, runnable on Windows, GNU/Linux, UNIX, and macOS.

Scenario

The benchmark simulates an online order processing system with multiple warehouses, districts, customers, and orders, representing a distributed database environment.

Five transaction types are defined, with the core performance metrics being New‑Order throughput (TPM) and latency.

New‑Order transaction throughput (TPM)

Latency

Environment preparation:

Operating System: Ubuntu 18.04.5 LTS

Container platform: KubeSphere V3.1.1

Database: RadonDB MySQL Kubernetes

Create Test Pod

$ kubectl run -i --tty --rm --image ubuntu test-shell bash
$ kubectl exec -ti test-shell -c test-shell /bin/bash
$ apt-get update

Install Tools

$ apt-get install make
$ apt-get install gcc
$ apt-get install git

Install MySQL client and development libraries:

$ apt-get install mysql-server
$ apt-get install libmysqlclient-dev

Clone and build TPCC‑MySQL:

$ git clone https://github.com/Percona-Lab/tpcc-mysql.git
$ cd tpcc-mysql/src
$ make

Data Preparation

In real tests, the number of warehouses should not be less than 100, and for SSD/PCIe SSD hardware, 1000 or more warehouses are recommended.

Create User

mysql> CREATE USER radondb@localhost IDENTIFIED BY 'mysql_password';
Query OK, 0 rows affected (0.00 sec)

mysql> grant all privileges on *.* to 'radondb'@'%' identified by 'mysql_password' with grant option;
Query OK, 0 rows affected, 1 warning (0.01 sec)

Create Database and Tables

Create the test database:

$ mysqladmin create tpcc1000 -h server_host -u mysql_user -p mysql_password

Load the provided schema files:

$ mysql -D tpcc1000 -h server_host -u mysql_user -p mysql_password < create_table.sql
$ mysql -D tpcc1000 -h server_host -u mysql_user -p mysql_password < add_fkey_idx.sql

Load Data

$ ./tpcc_load -h server_host -d tpcc1000 -u mysql_user -p mysql_password -w 20

Run TPC‑C Test

$ ./tpcc_start -h server_host -d tpcc1000 -u mysql_user -p mysql_password -w 20 -c 128 -r 120 -l 200 - > tpcc-output-log

Parameter explanation:

Generate Charts

Install gnuplot and generate performance graphs:

$ yum install -y gnuplot
$ cat log.conf | gnuplot

Summary

The TPC‑C results focus on two indicators:

Throughput (tpmC)

Measures the number of New‑Order transactions processed per minute while satisfying response‑time requirements for the other four transaction types; higher values are better.

Price/Performance (Price/tpmC)

Calculates the ratio of system price (U.S. list price) to throughput; lower values indicate better cost efficiency.