Skip to content

Latest commit

 

History

History
161 lines (118 loc) · 8.39 KB

DevelopmentGuide.md

File metadata and controls

161 lines (118 loc) · 8.39 KB
Raw

Development Guide

Setup

Make sure you have Amazon Corretto 8+ or Java 8+ installed.

Clone the AWS JDBC Driver repository:

git clone https://github.com/awslabs/aws-advanced-jdbc-wrapper.git

You can now make changes in the repository.

Building the AWS Advanced JDBC Driver

Navigate to project root:

cd aws-advanced-jdbc-wrapper

To build the AWS Advanced JDBC Driver without running the tests: Mac:

./gradlew build -x test

Windows:

gradlew build -x test

Mac:

./gradlew build

Windows:

gradlew build

Testing Overview

The AWS JDBC Driver uses the following tests to verify its correctness and performance on both JVM and GraalVM:

Tests Description
Unit tests Tests for AWS JDBC Driver correctness.
Failover integration tests Driver-specific tests for different reader and writer failover workflows using the Failover Connection Plugin.
Enhanced failure monitoring integration tests Driver-specific tests for the enhanced failure monitoring functionality using the Host Monitoring Connection Plugin.
AWS authentication integration tests Driver-specific tests for AWS authentication methods with the AWS Secrets Manager Plugin or the AWS IAM Authentication Plugin.
Connection plugin manager benchmarks The benchmarks subproject measures the overhead from executing JDBC method calls with multiple connection plugins enabled.

Extra Integration Tests

The AWS JDBC Driver repository also contains additional integration tests for external tools such as HikariCP, the Spring framework or Hibernate ORM.

The AWS JDBC Driver has been manually verified to work with database tools such as DBeaver.

Performance Tests

The JDBC Wrapper has 2 types of performance tests:

  • benchmarks measuring the AWS JDBC Driver's overhead when executing simple JDBC methods using the JMH microbenchmark framework
  • manually-triggered performance tests measuring the failover and enhanced failure monitoring plugins' performance under different configurations

AWS JDBC Driver Benchmarks

This diagram shows the benchmarks from running UPDATE queries using the AWS JDBC Driver with pgJDBC as the target driver. The baseline number represents running the same UPDATE queries with pgJDBC.

This diagram shows the benchmarks from running UPDATE queries using the AWS JDBC Driver with MySQL Connector/J as the target driver. The baseline number represents running the same UPDATE queries with MySQL Connector/J.

There are also specific benchmarks measuring the AWS JDBC Driver's pipelines. These benchmarks do not make actual connections to the databases and use simple test plugins. The goal of these benchmarks is to measure the overhead of using the AWS JDBC Driver with multiple plugins enabled.

See here for a more detailed performance breakdown.

Failover-specific Performance Tests

The diagrams in this section show the AWS JDBC Driver's failure detection performance with or without the Failover Connection Plugin under different settings. The performance tests share the following workflow:

  1. The AWS JDBC Driver executes an SQL query with a long execution time.
  2. After a network outage delay in milliseconds, the test triggers a network outage. Varying values for the network outage delay are tested, represented on the X axis of the diagrams below.
  3. Measures elapsed time between when the network outage and
    • when the AWS JDBC Driver detects the network failure if the Host Monitoring Connection Plugin is used, or
    • when the AWS JDBC Driver finishes the failover process if the Failover Connection Plugin is used.

This diagram shows the failover time with a 30-seconds socket timeout and different network outage delays. See here for a more detailed performance breakdown.

The following diagrams show how the AWS JDBC Driver performs under a more common failure detection setting versus a more aggressive setting.

Common Failure Detection Setting

Parameter Value
failoverTimeoutMs 120000
failureDetectionTime 30000
failureDetectionInterval 5000
failureDetectionCount 3

Aggressive Failure Detection Setting

Parameter Value
failoverTimeoutMs 120000
failureDetectionTime 6000
failureDetectionInterval 1000
failureDetectionCount 1

For more details on failure detection settings, see here. For more details on failover settings, see here. See here for a more detailed performance breakdown.

See here for a more detailed performance breakdown.

Read-Write Splitting and Internal Connection Pooling Performance Tests

The diagram in this section shows the AWS JDBC Driver's performance when using the read-write splitting plugin, with or without connection pooling enabled. This test sets up a large number of connections in parallel; the initial connection is to the writer but will be changed to a random reader instance. The test executes a long query many times to simulate heavy queries.

The average overhead time is measured as the average time difference between running the entire test with the read-write plugin and running the entire test without any plugins. The baseline overhead time is 0 because there are no plugins in this scenario and thus there is no plugin overhead.

Note that the given overhead values represent the total overhead of running the entire test workflow rather than the overhead of executing a single method. The test executes 1 writer query and 1 read-only query from 10 connections simultaneously.

See here for a more detailed performance breakdown.

Running the Tests

After building the AWS JDBC Driver you can now run the unit tests. This will also validate your environment is set up correctly.

Mac:

./gradlew test

Windows:

./gradlew test

Integration Tests

For more information on how to run the integration tests, please visit Integration Tests.

Sample Code

Connection Test Sample Code

Architecture

For more information on how the AWS Advanced JDBC Driver functions and how it is structured, please visit Architecture.