README.md

Triangle counting algorithm for large-scale directed graphs

algorithm configurations

algo: node_iter, node_iter_plus, edge_iter, edge_iter_plus, fast, fast_v2

sep: tab, space, comma

example for undirected graph

hdfs_host='hdfs_host'

/path_to_spark/bin/spark-submit
--class ch.ethz.ml.graph.runner.UndirectedRunner
--master yarn --deploy-mode cluster
--name "TriangleCount"
--driver-memory 2g
--num-executors 2
--executor-cores 2
--executor-memory 2g
--conf "spark.executor.extraJavaOptions=-XX:+UseG1GC -verbose:gc -XX:+PrintGCDateStamps -XX:+PrintGCTimeStamps"
graph-1.0-SNAPSHOT.jar mode:yarn-cluster
input:hdfs://${hdfs_host}:9000/path_to_graph/
output:hdfs://${hdfs_host}:9000/path_to_model/
cpDir:hdfs://${hdfs_host}:9000/path_to_cp/
partitionNum:4 storageLevel:DISK_ONLY
algo:fast_v2 sep:tab header:false

example for directed graph

hdfs_host='hdfs_host'

/path_to_spark/bin/spark-submit
--class ch.ethz.ml.graph.runner.DirectedRunner
--master yarn --deploy-mode cluster
--name "TriangleCount"
--driver-memory 2g
--num-executors 2
--executor-cores 2
--executor-memory 2g
--conf "spark.executor.extraJavaOptions=-XX:+UseG1GC -verbose:gc -XX:+PrintGCDateStamps -XX:+PrintGCTimeStamps"
graph-1.0-SNAPSHOT.jar mode:yarn-cluster
input:hdfs://${hdfs_host}:9000/path_to_graph/
output:hdfs://${hdfs_host}:9000/path_to_model/
cpDir:hdfs://${hdfs_host}:9000/path_to_cp/
partitionNum:4 storageLevel:DISK_ONLY
algo:fast_v2 sep:tab header:false