Status

dropwizard-hikaricp

This library provides a HikariCP integration for dropwizard, instead of using tomcat connection pool. It replaces the dropwizard-hibernate and dropwizard-db package, by overriding the classes DataSourceFactory and ManagedPooledDataSource.

DataSourceFactory overrides the original class by building a HikariConfig object, which is passed to the ManagedPooledDataSource. ManagedPooledDataSource extends HikariDataSource, instead of Tomcat's DataSourceProxy. It should work with existing dropwizard-hibernate settings, except for transaction isolation.

These are the available transaction isolation values. If not sure, leave it unset so it uses the default one.

• TRANSACTION_NONE
• TRANSACTION_READ_UNCOMMITTED
• TRANSACTION_READ_COMMITTED
• TRANSACTION_REPEATABLE_READ
• TRANSACTION_SERIALIZABLE

These are the supported versions of dropwizard:

Maven

The library is available at the maven central, so just add dependency to pom.xml:

<dependencies>
  <dependency>
    <groupId>com.github.mtakaki</groupId>
    <artifactId>dropwizard-hikaricp</artifactId>
    <version>2.0.20</version>
  </dependency>
</dependencies>

If the bundle is actually using HikariCP, you should be able to see this right in the beginning of the logs:

INFO  [2016-03-14 06:32:06,681] org.eclipse.jetty.util.log: Logging initialized @1894ms
INFO  [2016-03-14 06:32:08,675] com.zaxxer.hikari.HikariDataSource: hibernate - Starting...

If you don't see it, it means it's not using HikariCP.

Benchmark

Before looking into the benchmark numbers, I recommend reading HikariCP wiki explaining the differences between Hikari and Tomcat. There's also an extensive benchmarking produced by Nick Babcock, in which he talks about these results I've found, plus some very comprehensive tests.

The benchmark was ran on these specs:

• 1.4 GHz Intel Core i5 CPU
• Connection pool:
  • minimum size: 2
  • maximum size: 5
• 5000 requests
• 50 parallel clients

TomcatCP was used as it is and no additional setting was used to match to HikariCP's reliability.

Querying

Get single entry

Get multiple entries (5,000 records)

Inserting

Updating

Deleting