Modeling fcrepo4 time to run specs

io_perf_collect.sh

#!/bin/bash

# Performance testing normalization for disk related operations
#  Regression modeling for fcrepo performance (P) modeled by device write (W) and java startup time (J)
#  P ~ W + J


## CSV format for time output.  Will be munged by subsequent analysis
FORMAT_HEADER="wall_time,user_time,system_time"
FORMAT_STRING="%e,%U,%S"

# Collect duration data for storage device write operations.

## Run warmup
sh -c "dd if=/dev/zero of=ddfile bs=80k count=25000 status=none && sync"; rm ddfile

## K writes.
echo "K writes"
# Use -o k_writes.csv and --append options to write to file
/usr/bin/time -f ${FORMAT_STRING} sh -c "dd if=/dev/zero of=ddfile bs=8k count=250000 status=none && sync"; rm ddfile

## M writes.
echo "M writes"
# Use -o m_writes.csv and --append options to write to file
/usr/bin/time -f ${FORMAT_STRING} sh -c "dd if=/dev/zero of=ddfile bs=8000k count=250 status=none && sync"; rm ddfile

## G writes.
echo "G writes"
# Use -o g_writes.csv and --append options to write to file
/usr/bin/time -f ${FORMAT_STRING} sh -c "dd if=/dev/zero of=ddfile bs=1G count=2 status=none && sync"; rm ddfile


# Collect duration data for java startup times.

## Run warmup
bash -c "for i in {0..2}; do java -jar dummy_java.jar; done;"

## Time startups
echo "JVM starts"
/usr/bin/time -f ${FORMAT_STRING} bash -c "for i in {0..9}; do java -jar dummy_java.jar; done;"


# Collect duration data for fcrepo disk write tests. 

java org.junit.runner.JUnitCore junitfaq.AbstractResourceIT