Deploy Apache Ignite and Compare Its Performance with MySQL

Introduction

The article explains how to deploy, use, and benchmark Apache Ignite, a distributed in‑memory SQL database, and why it can be a valuable complement to traditional databases such as MySQL or Redis.

Why Choose Ignite?

Ignite implements the ANSI‑99 SQL standard, offers horizontal scalability, fault tolerance, and full support for distributed SQL joins. It can execute joins locally on each node without moving large amounts of data across the network, providing excellent performance for distributed workloads.

Case Overview

The example compares MySQL and Ignite performance within a single Spring Boot project. Different data sources are configured, and MyBatis is used to access each database. The setup also demonstrates how to add Ignite routing to an existing DB‑Router component.

Environment Setup

Docker Compose is used to start MySQL, Ignite, DBeaver, and ApacheBench. After pulling the images, Docker automatically creates the MySQL schema and launches an Ignite node.

Connection Configuration

Using DBeaver, you can connect to Ignite, verify the connection, create schemas and tables, and save the changes.

Project Structure

Application layer –

application-dev.yml

defines multiple data sources and loads them in

DataSourceConfig

.

Infrastructure layer – separate DAO packages for MySQL and Ignite.

Trigger layer – provides HTTP endpoints for testing Ignite and MySQL.

Data Source Creation

1. Maven Dependencies

<code>&lt;dependency&gt;
    &lt;groupId&gt;org.apache.ignite&lt;/groupId&gt;
    &lt;artifactId&gt;ignite-core&lt;/artifactId&gt;
    &lt;version&gt;2.15.0&lt;/version&gt;
&lt;/dependency&gt;

&lt;dependency&gt;
    &lt;groupId&gt;org.apache.ignite&lt;/groupId&gt;
    &lt;artifactId&gt;ignite-spring&lt;/artifactId&gt;
    &lt;version&gt;2.15.0&lt;/version&gt;
&lt;/dependency&gt;</code>

2. Ignite Configuration

<code>@Configuration
@MapperScan(basePackages = "cn.bugstack.xfg.dev.tech.infrastructure.ignite.dao", sqlSessionFactoryRef = "igniteSqlSessionFactory")
static class IgniteMyBatisConfig {
    @Bean("igniteDataSource")
    @ConfigurationProperties(prefix = "spring.ignite.datasource")
    public DataSource igniteDataSource(Environment env) {
        IgniteConfiguration cfg = new IgniteConfiguration();
        DataStorageConfiguration storage = new DataStorageConfiguration();
        DataRegionConfiguration region = new DataRegionConfiguration();
        region.setPersistenceEnabled(false);
        storage.setDefaultDataRegionConfiguration(region);
        cfg.setDataStorageConfiguration(storage);
        ConnectorConfiguration conn = new ConnectorConfiguration();
        conn.setIdleTimeout(60000);
        conn.setThreadPoolSize(100);
        cfg.setConnectorConfiguration(conn);
        return DataSourceBuilder.create()
                .url(env.getProperty("spring.ignite.datasource.url"))
                .driverClassName(env.getProperty("spring.ignite.datasource.driver-class-name"))
                .build();
    }

    @Bean("igniteSqlSessionFactory")
    public SqlSessionFactory igniteSqlSessionFactory(DataSource igniteDataSource) throws Exception {
        SqlSessionFactoryBean bean = new SqlSessionFactoryBean();
        bean.setDataSource(igniteDataSource);
        bean.setMapperLocations(new PathMatchingResourcePatternResolver()
                .getResources("classpath:/mybatis/mapper/ignite/*.xml"));
        return bean.getObject();
    }
}
</code>

3. MySQL Configuration

<code>@Configuration
@MapperScan(basePackages = "cn.bugstack.xfg.dev.tech.infrastructure.mysql.dao", sqlSessionFactoryRef = "mysqlSqlSessionFactory")
static class MysqlMyBatisConfig {
    @Bean("mysqlDataSource")
    @ConfigurationProperties(prefix = "spring.mysql.datasource")
    public DataSource mysqlDataSource(Environment env) {
        return DataSourceBuilder.create()
                .url(env.getProperty("spring.mysql.datasource.url"))
                .driverClassName(env.getProperty("spring.mysql.datasource.driver-class-name"))
                .build();
    }

    @Bean("mysqlSqlSessionFactory")
    public SqlSessionFactory mysqlSqlSessionFactory(DataSource mysqlDataSource) throws Exception {
        SqlSessionFactoryBean bean = new SqlSessionFactoryBean();
        bean.setDataSource(mysqlDataSource);
        bean.setMapperLocations(new PathMatchingResourcePatternResolver()
                .getResources("classpath:/mybatis/mapper/mysql/*.xml"));
        return bean.getObject();
    }
}
</code>

Performance Testing

Two HTTP controllers expose endpoints for load testing with

ab

. The same set of commands is used for both Ignite and MySQL.

Ignite Load Test

Initialize ID:

ab -c 1 -n 1 http://127.0.0.1:8091/api/ignite/start

Insert 50,000 rows:

ab -c 20 -n 50000 http://127.0.0.1:8091/api/ignite/insert

Read 1,000 rows:

ab -c 20 -n 1000 http://127.0.0.1:8091/api/ignite/cacheData

Query by OrderId (with index):

ab -c 20 -n 1000 http://127.0.0.1:8091/api/ignite/selectByOrderId

MySQL Load Test

Initialize ID:

ab -c 1 -n 1 http://127.0.0.1:8091/api/ignite/start

Insert 50,000 rows:

ab -c 20 -n 50000 http://127.0.0.1:8091/api/ignite/insert

Read 1,000 rows:

ab -c 20 -n 1000 http://127.0.0.1:8091/api/ignite/cacheData

Query by OrderId (with index):

ab -c 20 -n 1000 http://127.0.0.1:8091/api/ignite/selectByOrderId

Conclusion

Ignite shows a slight performance advantage over MySQL, confirming that a pure in‑memory database can be faster for workloads that benefit from memory computing while keeping the existing MySQL schema unchanged.