Skip to content

High performance reactive PostgreSQL client written in Java

License

Notifications You must be signed in to change notification settings

BillyYccc/reactive-pg-client

 
 

Repository files navigation

Reactive Postgres Client

Build Status

  • Simple API focusing on scalability and low overhead.

  • Reactive and non blocking which able to handle many database connections with a single thread.

  • Ranked #1 in the TechEmpower Benchmark Round 15 Single query benchmark.

This project has evolved to become the Reactive SQL Client that provides support for PostgreSQL and MySQL.

This project will remain maintained for bug fixes.

Features

  • Event driven

  • Lightweight

  • Built-in connection pooling

  • Prepared queries caching

  • Publish / subscribe using Postgres LISTEN/NOTIFY

  • Batch and cursor support

  • Row streaming

  • java.util.stream.Collector row set transformation

  • Command pipeling

  • RxJava 1 and RxJava 2

  • Direct memory to object without unnecessary copies

  • Java 8 Date and Time

  • SSL/TLS

  • Unix domain socket

  • HTTP/1.x CONNECT, SOCKS4a or SOCKS5 proxy

  • Request cancellation

Usage

Latest release is 0.11.3.

To use the client, add the following dependency to the dependencies section of your build descriptor:

  • Maven (in your pom.xml) for Vert.x 3.7.x:

<dependency>
  <groupId>io.reactiverse</groupId>
  <artifactId>reactive-pg-client</artifactId>
  <version>0.11.3</version>
</dependency>
  • Gradle (in your build.gradle file) for Vert.x 3.7.x:

dependencies {
  compile 'io.reactiverse:reactive-pg-client:0.11.3'
}

If you are using Vertx 3.5.x you should use instead 0.10.9

Then the code is quite straightforward:

// Pool options
PgPoolOptions options = new PgPoolOptions()
  .setPort(5432)
  .setHost("the-host")
  .setDatabase("the-db")
  .setUser("user")
  .setPassword("secret")
  .setMaxSize(5);

// Create the client pool
PgPool client = PgClient.pool(options);

// A simple query
client.query("SELECT * FROM users WHERE id='julien'", ar -> {
  if (ar.succeeded()) {
    PgResult<Row> result = ar.result();
    System.out.println("Got " + result.size() + " results ");
  } else {
    System.out.println("Failure: " + ar.cause().getMessage());
  }

  // Now close the pool
  client.close();
});

Integration / Usages

  • vertx-jooq: writing type safe queries and let them run at high speed using jOOQ, vertx and the reactive-pg-client.

  • Micronaut: supports reactive and non-blocking client to connect to Postgres using reactive-pg-client, allowing to handle many database connections with a single thread.

Javadoc

Pipelining

This client supports pipelining requests to the database which can give a significant performance improvement depending on the latency to the database and the type of queries your application is doing.

100µs latency
Figure 1. 100µs latency
1ms latency
Figure 2. 1ms latency

Such results have been produced using this benchmark.

Warning
The two results are not normalized, the 100µs latency executes the 5000 queries in about 300ms, the 1ms latency executes the 5000 queries in about 13 seconds.

Supported Data Types

The Reactive Postgres Client currently supports the following data types

_ Value Array

Postgres

Java

Supported

JAVA

Supported

BOOLEAN

j.l.Boolean

j.l.Boolean[]

INT2

j.l.Short

j.l.Short[]

INT4

j.l.Integer

j.l.Integer[]

INT8

j.l.Long

j.l.Long[]

FLOAT4

j.l.Float

j.l.Float[]

FLOAT8

j.l.Double

j.l.Double[]

CHAR

j.l.Character

j.l.Character[]

VARCHAR

j.l.String

j.l.String[]

TEXT

j.l.String

j.l.String[]

ENUM

j.l.String

j.l.String[]

NAME

j.l.String

j.l.String[]

SERIAL2

j.l.Short

invalid type

SERIAL4

j.l.Integer

invalid type

SERIAL8

j.l.Long

invalid type

NUMERIC

i.r.p.data.Numeric

i.r.p.data.Numeric[]

UUID

j.u.UUID

j.u.UUID[]

DATE

j.t.LocalDate

j.t.LocalDate[]

TIME

j.t.LocalTime

j.t.LocalTime[]

TIMETZ

j.t.OffsetTime

j.t.OffsetTime[]

TIMESTAMP

j.t.LocalDateTime

j.t.LocalDateTime[]

TIMESTAMPTZ

j.t.OffsetDateTime

j.t.OffsetDateTime[]

INTERVAL

i.r.p.data.Interval

i.r.p.data.Interval[]

BYTEA

i.v.c.b.Buffer

i.v.c.b.Buffer[]

JSON

i.r.p.data.Json

i.r.p.data.Json[]

JSONB

i.r.p.data.Json

i.r.p.data.Json[]

POINT

i.r.p.data.Point

i.r.p.data.Point[]

LINE

i.r.p.data.Line

i.r.p.data.Line[]

LSEG

i.r.p.data.LineSegment

i.r.p.data.LineSegment[]

BOX

i.r.p.data.Box

i.r.p.data.Box[]

PATH

i.r.p.data.Path

i.r.p.data.Path[]

POLYGON

i.r.p.data.Polygon

i.r.p.data.Polygon[]

CIRCLE

i.r.p.data.Circle

i.r.p.data.Circle[]

UNKNOWN

j.l.String

j.l.String[]

The following types

MONEY, BIT, VARBIT, MACADDR, INET, CIDR, MACADDR8, XML, HSTORE, OID, VOID, TSQUERY, TSVECTOR

are not implemented yet (PR are welcome).

Snapshots

License

Apache License - Version 2.0

Developers

Testing

Out of the box, the test suite runs an embedded Postgres by default.

You can change the version of the embedded Postgres by passing a property embedded.postgres.version like this:

> mvn test -Dembedded.postgres.version=9.6

The following versions of embedded Postgres are supported:

  • 9.6

  • 10.6 (default)

  • 11.1 (not supported on Linux)

Testing with an external database

You can run tests with an external database:

  • the script docker/postgres/resources/create-postgres.sql creates the test data

  • the TLSTest expects the database to be configured with SSL with docker/postgres/resources/server.key / docker/postgres/resources/server.cert`

You need to add some properties for testing:

> mvn test -Dconnection.uri=postgres://$username:$password@$host:$port/$database -Dtls.connection.uri=postgres://$username:$password@$host:$port/$database -Dunix.socket.directory=$path
  • connection.uri(mandatory): configure the client to connect the specified database

  • tls.connection.uri(mandatory): configure the client to run TLSTest with the specified Postgres with SSL enabled

  • unix.socket.directory(optional): the single unix socket directory(multiple socket directories are not supported) to test Unix domain socket with a specified database, domain socket tests will be skipped if this property is not specified (Note: Make sure you can access the unix domain socket with this directory under your host machine)

  • unix.socket.port(optional): unix socket file is named .s.PGSQL.nnnn and nnnn is the server’s port number, this property is mostly used when you test with Docker, when you publish your Postgres container port other than 5432 in your host but Postgres may actually listen on a different port in the container, you will then need this property to help you connect the Postgres with Unix domain socket

Testing with Docker

Run the Postgres containers with docker-compose:

> cd docker/postgres
> docker-compose up --build -V

Run tests:

> mvn test -Dconnection.uri=postgres://$username:$password@$host:$port/$database -Dtls.connection.uri=postgres://$username:$password@$host:$port/$database -Dunix.socket.directory=$path -Dunix.socket.port=$port

Documentation

The online and published documentation is in /docs and is served by GitHub pages with Jekyll.

You can find the actual guide source in [src/main/docs/index.md](src/main/docs/index.md). At compilation time, this source generates the jekyll/guide/java/index.md.

The current documentation is in /jekyll and can be preview using Docker and your browser

  • generate the documentation

    • mvn compile to generate jekyll/guide/java/index.md

    • mvn site to generate the javadoc in jekyll/apidocs

  • run Jekyll

    • cd jekyll

    • docker-compose up

  • open your browser at http://localhost:4000

About

High performance reactive PostgreSQL client written in Java

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Java 89.7%
  • Kotlin 8.1%
  • PLpgSQL 2.0%
  • Other 0.2%