mapreduce/src/benchmarks/gridmix/gridmix-env

#!/usr/bin/env bash


## Environment configuration
# Hadoop installation
# set var only if it has not already been set externally
if [ -z "${HADOOP_HOME}" ] ; then
  export HADOOP_HOME=
fi
# Base directory for gridmix install
# set var only if it has not already been set externally
if [ -z "${GRID_MIX_HOME}" ] ; then
  export GRID_MIX_HOME=${GRID_DIR}
fi
# Hadoop example jar
# set var only if it has not already been set externally
if [ -z "${EXAMPLE_JAR}" ] ; then
  export EXAMPLE_JAR="${HADOOP_HOME}/hadoop-*examples.jar"
fi
# Hadoop test jar
# set var only if it has not already been set externally
if [ -z "${APP_JAR}" ] ; then
  export APP_JAR="${HADOOP_HOME}/hadoop-*test.jar"
fi
# Hadoop streaming jar
# set var only if it has not already been set externally
if [ -z "${STREAM_JAR}" ] ; then
  export STREAM_JAR="${HADOOP_HOME}/contrib/streaming/hadoop-*streaming.jar"
fi
# Location on default filesystem for writing gridmix data (usually HDFS)
# Default: /gridmix/data
# set var only if it has not already been set externally
if [ -z "${GRID_MIX_DATA}" ] ; then
  export GRID_MIX_DATA=/gridmix/data
fi
# Location of executables in default filesystem (usually HDFS)
# Default: /gridmix/programs
# set var only if it has not already been set externally
if [ -z "${GRID_MIX_PROG}" ] ; then
  export GRID_MIX_PROG=/gridmix/programs
fi

## Data sources
# Variable length key, value compressed SequenceFile
export VARCOMPSEQ=${GRID_MIX_DATA}/WebSimulationBlockCompressed
# Fixed length key, value compressed SequenceFile
export FIXCOMPSEQ=${GRID_MIX_DATA}/MonsterQueryBlockCompressed
# Variable length key, value uncompressed Text File
export VARINFLTEXT=${GRID_MIX_DATA}/SortUncompressed
# Fixed length key, value compressed Text File
export FIXCOMPTEXT=${GRID_MIX_DATA}/EntropySimulationCompressed

## Job sizing
export NUM_OF_LARGE_JOBS_FOR_ENTROPY_CLASS=5
export NUM_OF_LARGE_JOBS_PER_CLASS=3
export NUM_OF_MEDIUM_JOBS_PER_CLASS=20
export NUM_OF_SMALL_JOBS_PER_CLASS=40

export NUM_OF_REDUCERS_FOR_LARGE_JOB=370
export NUM_OF_REDUCERS_FOR_MEDIUM_JOB=170
export NUM_OF_REDUCERS_FOR_SMALL_JOB=15

## Throttling
export INTERVAL_BETWEEN_SUBMITION=20

## Hod
#export HOD_OPTIONS=""

export CLUSTER_DIR_BASE=$GRID_MIX_HOME/CLUSTER_DIR_BASE
export HOD_CONFIG=
export ALL_HOD_OPTIONS="-c ${HOD_CONFIG} ${HOD_OPTIONS}"
export SMALL_JOB_HOD_OPTIONS="$ALL_HOD_OPTIONS -n 5"
export MEDIUM_JOB_HOD_OPTIONS="$ALL_HOD_OPTIONS -n 50"
export LARGE_JOB_HOD_OPTIONS="$ALL_HOD_OPTIONS -n 100"