Flink performance tests
Switch branches/tags
Nothing to show
Clone or download
Pull request Compare This branch is 13 commits behind project-flink:master.

README.md

Apache Flink: Performance and Testing

Flink performance tests

Add

WordCount WordCount NoComb

K-Means

  • low dimensional (3 dimensions k =20)
  • high dimensional (1000 dimensions, k =200)

TPC-H with two joins and aggregation (Q3 if suitable)

Connected components

PageRank

Contains (for now) one large test job that tests all Flink components.

Background information

This code originated from the Stratosphere.eu project. It is now called Apache Flink (incubating).

Original issue for planning this feature: https://github.com/stratosphere/stratosphere/issues/379

Preparation

Generate TPC-H Data

Download and build the TPC-H generate tool using ./prepareTPCH.sh.

Generate data using ./generateTPCH.sh

Generate Special Data

  1. Generate Avro File

This command will generate an Avro file from the original TPC-H orders file. (This step is not necessary when running the test job locally.)

java -cp target/testjob-0.1-SNAPSHOT.jar eu.stratosphere.test.testPlan.LargeTestPlan '/input/orders.tbl' '/output/orders.avro'
  1. Generate Sequence File
java -cp target/testjob-0.1-SNAPSHOT.jar eu.stratosphere.test.testPlan.SequenceFileGenerator SeqOut 1000000 15

(45 MB)

Execute Plan

Execute Plan with LocalExecutor

The following parameters need to be passed to the main()-Method of LargeTestPlan (in that order).

Parameter Description Example
customer path to TPC-H file file:///.../customer.tbl
lineitem path to TPC-H file file:///.../lineitem.tbl
nation path to TPC-H file file:///.../nation.tbl
orders path to TPC-H file file:///.../orders.tbl
region path to TPC-H file file:///.../region.tbl
orderAvroFile path to generated Avro file file:///.../orders.avro
sequenceFileInput path to generated Sequence file file:///.../seqfile
outputTableDirectory path to test output directory file:///.../directory/
maxBulkIterations max bulk iterations (Test 10), should be greater than number of orders 1000
ordersPath ordinary path to TPC-H order file /.../orders.tbl
outputAccumulatorsPath ordinary path where accumulator results get stored /.../accumulators.txt
outputKeylessReducerPath ordinary path where AllReduce count results get stored /.../allreduce.txt
outputOrderAvroPath ordinary path where the generated Avro file gets stored /.../orders.avro

Execute Plan on Cluster

The following parameters need to be passed to the Job (in that order). The Avro file and the Sequence file need to be generated previously.

Parameter Description Example
customer path to TPC-H file file:///.../customer.tbl
lineitem path to TPC-H file file:///.../lineitem.tbl
nation path to TPC-H file file:///.../nation.tbl
orders path to TPC-H file file:///.../orders.tbl
region path to TPC-H file file:///.../region.tbl
orderAvroFile path to generated Avro file file:///.../orders.avro
sequenceFileInput path to generated Sequence file file:///.../seqfile
outputTableDirectory path to test output directory file:///.../directory/
maxBulkIterations max bulk iterations (Test 10), should be greater than number of orders 1000