SparkUCX is a high performance ShuffleManager plugin for Apache Spark, that uses RDMA and other high performance transports that are supported by UCX, to perform Shuffle data transfers in Spark jobs.
This open-source project is developed, maintained and supported by the UCF consortium.
- Apache Spark 2.3/2.4/3.0
- Java 8+
- Installed UCX of version 1.10+, and UCX supported transport hardware.
Please use the "Releases" page to download SparkUCX jar file
for your spark version (e.g. spark-ucx-1.0-for-spark-2.4.0-jar-with-dependencies.jar).
Put SparkUCX jar file in $SPARK_UCX_HOME on all the nodes in your cluster.
If you would like to build the project yourself, please refer to the "Build" section below.
Ucx binaries must be in Spark classpath on every Spark Master and Worker. It can be obtained by installing the latest version from Ucx release page
Provide Spark the location of the SparkUCX plugin jars and ucx shared binaries by using the extraClassPath option.
spark.driver.extraClassPath $SPARK_UCX_HOME/spark-ucx-1.0-for-spark-2.4.0-jar-with-dependencies.jar:$UCX_PREFIX/lib
spark.executor.extraClassPath $SPARK_UCX_HOME/spark-ucx-1.0-for-spark-2.4.0-jar-with-dependencies.jar:$UCX_PREFIX/lib
To enable the SparkUCX Shuffle Manager plugin, add the following configuration property to spark (e.g. in $SPARK_HOME/conf/spark-defaults.conf):
spark.shuffle.manager org.apache.spark.shuffle.UcxShuffleManager
For spark-3.0 version add SparkUCX ShuffleIO plugin:
spark.shuffle.sort.io.plugin.class org.apache.spark.shuffle.compat.spark_3_0.UcxLocalDiskShuffleDataIO
Building the SparkUCX plugin requires Apache Maven and Java 8+ JDK
Build instructions:
% git clone https://github.com/openucx/sparkucx
% cd sparkucx
% mvn -DskipTests clean package -Pspark-2.4
SparkUCX plugin is built to provide the best performance out-of-the-box, and provides multiple configuration options to further tune SparkUCX per-job. For more information on how to setup HiBench benchmark and reproduce results, please refer to Accelerated Apache SparkUCX 2.4/3.0 cluster deployment.