New Features, Benchmarks, and Install Guide for Alibaba Dragonwell JDK 8.0.0‑GA

Dragonwell 8.0.0‑GA Release Overview

Alibaba Dragonwell JDK 8.0.0‑GA was released in April 2023. The GA build synchronizes the upstream OpenJDK 8u212‑b04 source, incorporating the latest security fixes and patches. In addition to upstream updates, Dragonwell fixes several bugs observed in large‑scale Alibaba Java workloads and provides a default security certificate that will be maintained continuously.

Key Features

JWarmUp – Reducing Warm‑up Latency

JWarmUp addresses the long warm‑up period of JVMs in high‑concurrency scenarios (e.g., Alibaba’s Double‑11 shopping festival). Instead of using manually crafted warm‑up data, Dragonwell collects real execution hotspots from previous runs and re‑uses this information to pre‑warm the JVM.

Hotspot collection uses real workload data, improving compilation quality.

Hot methods are compiled immediately upon class loading, accelerating the warm‑up process.

Benchmark on specjvm2008 shows a performance increase for the xml.validation test case:

# Standard OpenJDK score on xml.validation: 268.07 ops/m
Score on xml.validation (Dragonwell): 294.95 ops/m

This represents roughly a 10 % improvement and the feature is being proposed as a JEP for upstream OpenJDK.

Java Flight Recorder (JFR) Enhancements

JFR records runtime data with minimal overhead, enabling detailed analysis via Java Mission Control (JMC). Dragonwell integrates JFR into OpenJDK 8u and contributes backport work to the OpenJDK 8 mainline. In addition to upstream capabilities, Dragonwell adds early‑access enhancements that are not yet merged upstream.

ElasticHeap – Adaptive Heap Management

ElasticHeap is a G1‑GC‑based dynamic heap elasticity feature that returns unused memory to the operating system without requiring Full GC pauses.

Automatic adjustment based on allocation speed and GC pressure.

Manual control via jcmd or MXBean commands, supporting whole‑heap or generation‑specific limits.

During peak traffic, ElasticHeap expands the heap under high GC pressure and promptly releases memory when load subsides, achieving a 20‑30 % reduction in physical memory usage during low‑load periods.

Installation and Usage

Dragonwell 8 supports Linux x86‑64.

Binary tarball

Download the pre‑compiled tarball from the GitHub releases page:

https://github.com/alibaba/dragonwell8/releases

Extract the tarball to the desired installation directory.

YUM installation

Dragonwell packages are available in the Alibaba Cloud Linux 2 YUM repository, which is compatible with Aliyun Linux 17.1, RHEL 7, and CentOS 7.

Install via YUM:

sudo yum install -y java-1.8.0-alibaba-dragonwell

After installation, configure the environment:

export JAVA_HOME=/opt/alibaba/java-1.8.0-alibaba-dragonwell-8.0.0.212.b04-1.al7
export PATH=$JAVA_HOME/bin:$PATH

If the YUM repository is not already configured, add the following repo file to /etc/yum.repos.d/:

[plus]
name=AliYun-2.1903-Plus-mirrors.aliyun.com
baseurl=http://mirrors.aliyun.com/alinux/2.1903/plus/$basearch/
gpgcheck=1
gpgkey=http://mirrors.aliyun.com/alinux/RPM-GPG-KEY-ALIYUN

References

Release notes (Chinese): https://github.com/alibaba/dragonwell8/wiki/%E9%98%BF%E9%87%8C%E5%B7%B4%E5%B7%B4Dragonwell8%E5%8F%91%E5%B8%83%E8%AF%B4%E6%98%8E

Talk slides (English): http://www.cs.utah.edu/cloudruntimes/slides/xiaoming.pdf