rocksdb/tools/benchmark.sh
Gunnar Kudrjavets 697fab820a Updates to RocksDB subcompaction benchmarking script
Summary: Set of updates to the subcompaction benchmark script which are based on our internal discussions. The intent behind the changes is to make sure that the scripts will correctly reflect how we're doing the actual benchmarking.

Test Plan: Tested by exercising the full set of compaction benchmarks and validating the execution and consistency of results.

Reviewers: MarkCallaghan, sdong, yhchiang

Reviewed By: yhchiang

Subscribers: andrewkr, dhruba

Differential Revision: https://reviews.facebook.net/D55461
2016-03-14 23:09:04 -07:00

502 lines
16 KiB
Bash
Executable File

#!/bin/bash
# REQUIRE: db_bench binary exists in the current directory
if [ $# -ne 1 ]; then
echo -n "./benchmark.sh [bulkload/fillseq/overwrite/filluniquerandom/"
echo "readrandom/readwhilewriting/readwhilemerging/updaterandom/"
echo "mergerandom/randomtransaction/compact]"
exit 0
fi
# Make it easier to run only the compaction test. Getting valid data requires
# a number of iterations and having an ability to run the test separately from
# rest of the benchmarks helps.
if [ "$COMPACTION_TEST" == "1" -a "$1" != "universal_compaction" ]; then
echo "Skipping $1 because it's not a compaction test."
exit 0
fi
# size constants
K=1024
M=$((1024 * K))
G=$((1024 * M))
if [ -z $DB_DIR ]; then
echo "DB_DIR is not defined"
exit 0
fi
if [ -z $WAL_DIR ]; then
echo "WAL_DIR is not defined"
exit 0
fi
output_dir=${OUTPUT_DIR:-/tmp/}
if [ ! -d $output_dir ]; then
mkdir -p $output_dir
fi
# all multithreaded tests run with sync=1 unless
# $DB_BENCH_NO_SYNC is defined
syncval="1"
if [ ! -z $DB_BENCH_NO_SYNC ]; then
echo "Turning sync off for all multithreaded tests"
syncval="0";
fi
num_threads=${NUM_THREADS:-16}
mb_written_per_sec=${MB_WRITE_PER_SEC:-0}
# Only for tests that do range scans
num_nexts_per_seek=${NUM_NEXTS_PER_SEEK:-10}
cache_size=${CACHE_SIZE:-$((1 * G))}
duration=${DURATION:-0}
num_keys=${NUM_KEYS:-$((1 * G))}
key_size=20
value_size=${VALUE_SIZE:-400}
block_size=${BLOCK_SIZE:-8192}
const_params="
--db=$DB_DIR \
--wal_dir=$WAL_DIR \
--disable_data_sync=0 \
\
--num=$num_keys \
--num_levels=6 \
--key_size=$key_size \
--value_size=$value_size \
--block_size=$block_size \
--cache_size=$cache_size \
--cache_numshardbits=6 \
--compression_type=snappy \
--min_level_to_compress=3 \
--compression_ratio=0.5 \
--level_compaction_dynamic_level_bytes=true \
--bytes_per_sync=$((8 * M)) \
--cache_index_and_filter_blocks=0 \
--benchmark_write_rate_limit=$(( 1024 * 1024 * $mb_written_per_sec )) \
\
--hard_rate_limit=3 \
--rate_limit_delay_max_milliseconds=1000000 \
--write_buffer_size=$((128 * M)) \
--max_write_buffer_number=8 \
--target_file_size_base=$((128 * M)) \
--max_bytes_for_level_base=$((1 * G)) \
\
--verify_checksum=1 \
--delete_obsolete_files_period_micros=$((60 * M)) \
--max_grandparent_overlap_factor=8 \
--max_bytes_for_level_multiplier=8 \
\
--statistics=0 \
--stats_per_interval=1 \
--stats_interval_seconds=60 \
--histogram=1 \
\
--memtablerep=skip_list \
--bloom_bits=10 \
--open_files=-1"
l0_config="
--level0_file_num_compaction_trigger=4 \
--level0_slowdown_writes_trigger=12 \
--level0_stop_writes_trigger=20"
if [ $duration -gt 0 ]; then
const_params="$const_params --duration=$duration"
fi
params_w="$const_params $l0_config --max_background_compactions=16 --max_background_flushes=7"
params_bulkload="$const_params --max_background_compactions=16 --max_background_flushes=7 \
--level0_file_num_compaction_trigger=$((10 * M)) \
--level0_slowdown_writes_trigger=$((10 * M)) \
--level0_stop_writes_trigger=$((10 * M))"
#
# Tune values for level and universal compaction.
# For universal compaction, these level0_* options mean total sorted of runs in
# LSM. In level-based compaction, it means number of L0 files.
#
params_level_compact="$const_params \
--max_background_flushes=4 \
--level0_file_num_compaction_trigger=4 \
--level0_slowdown_writes_trigger=16 \
--level0_stop_writes_trigger=20"
params_univ_compact="$const_params \
--max_background_flushes=4 \
--level0_file_num_compaction_trigger=8 \
--level0_slowdown_writes_trigger=16 \
--level0_stop_writes_trigger=20"
function summarize_result {
test_out=$1
test_name=$2
bench_name=$3
# Note that this function assumes that the benchmark executes long enough so
# that "Compaction Stats" is written to stdout at least once. If it won't
# happen then empty output from grep when searching for "Sum" will cause
# syntax errors.
uptime=$( grep ^Uptime\(secs $test_out | tail -1 | awk '{ printf "%.0f", $2 }' )
stall_time=$( grep "^Cumulative stall" $test_out | tail -1 | awk '{ print $3 }' )
stall_pct=$( grep "^Cumulative stall" $test_out| tail -1 | awk '{ print $5 }' )
ops_sec=$( grep ^${bench_name} $test_out | awk '{ print $5 }' )
mb_sec=$( grep ^${bench_name} $test_out | awk '{ print $7 }' )
lo_wgb=$( grep "^ L0" $test_out | tail -1 | awk '{ print $8 }' )
sum_wgb=$( grep "^ Sum" $test_out | tail -1 | awk '{ print $8 }' )
sum_size=$( grep "^ Sum" $test_out | tail -1 | awk '{ printf "%.1f", $3 / 1024.0 }' )
wamp=$( echo "scale=1; $sum_wgb / $lo_wgb" | bc )
wmb_ps=$( echo "scale=1; ( $sum_wgb * 1024.0 ) / $uptime" | bc )
usecs_op=$( grep ^${bench_name} $test_out | awk '{ printf "%.1f", $3 }' )
p50=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.1f", $3 }' )
p75=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.1f", $5 }' )
p99=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $7 }' )
p999=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $9 }' )
p9999=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $11 }' )
echo -e "$ops_sec\t$mb_sec\t$sum_size\t$lo_wgb\t$sum_wgb\t$wamp\t$wmb_ps\t$usecs_op\t$p50\t$p75\t$p99\t$p999\t$p9999\t$uptime\t$stall_time\t$stall_pct\t$test_name" \
>> $output_dir/report.txt
}
function run_bulkload {
# This runs with a vector memtable and the WAL disabled to load faster. It is still crash safe and the
# client can discover where to restart a load after a crash. I think this is a good way to load.
echo "Bulk loading $num_keys random keys"
cmd="./db_bench --benchmarks=fillrandom \
--use_existing_db=0 \
--disable_auto_compactions=1 \
--sync=0 \
$params_bulkload \
--threads=1 \
--memtablerep=vector \
--disable_wal=1 \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/benchmark_bulkload_fillrandom.log"
echo $cmd | tee $output_dir/benchmark_bulkload_fillrandom.log
eval $cmd
summarize_result $output_dir/benchmark_bulkload_fillrandom.log bulkload fillrandom
echo "Compacting..."
cmd="./db_bench --benchmarks=compact \
--use_existing_db=1 \
--disable_auto_compactions=1 \
--sync=0 \
$params_w \
--threads=1 \
2>&1 | tee -a $output_dir/benchmark_bulkload_compact.log"
echo $cmd | tee $output_dir/benchmark_bulkload_compact.log
eval $cmd
}
#
# Parameter description:
#
# $1 - 1 if I/O statistics should be collected.
# $2 - compaction type to use (level=0, universal=1).
# $3 - number of subcompactions.
# $4 - number of maximum background compactions.
#
function run_manual_compaction_worker {
# This runs with a vector memtable and the WAL disabled to load faster.
# It is still crash safe and the client can discover where to restart a
# load after a crash. I think this is a good way to load.
echo "Bulk loading $num_keys random keys for manual compaction."
fillrandom_output_file=$output_dir/benchmark_man_compact_fillrandom_$3.log
man_compact_output_log=$output_dir/benchmark_man_compact_$3.log
if [ "$2" == "1" ]; then
extra_params=$params_univ_compact
else
extra_params=$params_level_compact
fi
# Make sure that fillrandom uses the same compaction options as compact.
cmd="./db_bench --benchmarks=fillrandom \
--use_existing_db=0 \
--disable_auto_compactions=0 \
--sync=0 \
$extra_params \
--threads=$num_threads \
--compaction_measure_io_stats=$1 \
--compaction_style=$2 \
--subcompactions=$3 \
--memtablerep=vector \
--disable_wal=1 \
--max_background_compactions=$4 \
--seed=$( date +%s ) \
2>&1 | tee -a $fillrandom_output_file"
echo $cmd | tee $fillrandom_output_file
eval $cmd
summarize_result $fillrandom_output_file man_compact_fillrandom_$3 fillrandom
echo "Compacting with $3 subcompactions specified ..."
# This is the part we're really interested in. Given that compact benchmark
# doesn't output regular statistics then we'll just use the time command to
# measure how long this step takes.
cmd="{ \
time ./db_bench --benchmarks=compact \
--use_existing_db=1 \
--disable_auto_compactions=0 \
--sync=0 \
$extra_params \
--threads=$num_threads \
--compaction_measure_io_stats=$1 \
--compaction_style=$2 \
--subcompactions=$3 \
--max_background_compactions=$4 \
;}
2>&1 | tee -a $man_compact_output_log"
echo $cmd | tee $man_compact_output_log
eval $cmd
# Can't use summarize_result here. One way to analyze the results is to run
# "grep real" on the resulting log files.
}
function run_univ_compaction {
# Always ask for I/O statistics to be measured.
io_stats=1
# Values: kCompactionStyleLevel = 0x0, kCompactionStyleUniversal = 0x1.
compaction_style=1
# Define a set of benchmarks.
subcompactions=(1 2 4 8 16)
max_background_compactions=(16 16 8 4 2)
i=0
total=${#subcompactions[@]}
# Execute a set of benchmarks to cover variety of scenarios.
while [ "$i" -lt "$total" ]
do
run_manual_compaction_worker $io_stats $compaction_style ${subcompactions[$i]} \
${max_background_compactions[$i]}
((i++))
done
}
function run_fillseq {
# This runs with a vector memtable. WAL can be either disabled or enabled
# depending on the input parameter (1 for disabled, 0 for enabled). The main
# benefit behind disabling WAL is to make loading faster. It is still crash
# safe and the client can discover where to restart a load after a crash. I
# think this is a good way to load.
# Make sure that we'll have unique names for all the files so that data won't
# be overwritten.
if [ $1 == 1 ]; then
log_file_name=$output_dir/benchmark_fillseq.wal_disabled.v${value_size}.log
test_name=fillseq.wal_disabled.v${value_size}
else
log_file_name=$output_dir/benchmark_fillseq.wal_enabled.v${value_size}.log
test_name=fillseq.wal_enabled.v${value_size}
fi
echo "Loading $num_keys keys sequentially"
cmd="./db_bench --benchmarks=fillseq \
--use_existing_db=0 \
--sync=0 \
$params_w \
--min_level_to_compress=0 \
--threads=1 \
--memtablerep=vector \
--disable_wal=$1 \
--seed=$( date +%s ) \
2>&1 | tee -a $log_file_name"
echo $cmd | tee $log_file_name
eval $cmd
# The constant "fillseq" which we pass to db_bench is the benchmark name.
summarize_result $log_file_name $test_name fillseq
}
function run_change {
operation=$1
echo "Do $num_keys random $operation"
out_name="benchmark_${operation}.t${num_threads}.s${syncval}.log"
cmd="./db_bench --benchmarks=$operation \
--use_existing_db=1 \
--sync=$syncval \
$params_w \
--threads=$num_threads \
--merge_operator=\"put\" \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/${out_name}"
echo $cmd | tee $output_dir/${out_name}
eval $cmd
summarize_result $output_dir/${out_name} ${operation}.t${num_threads}.s${syncval} $operation
}
function run_filluniquerandom {
echo "Loading $num_keys unique keys randomly"
cmd="./db_bench --benchmarks=filluniquerandom \
--use_existing_db=0 \
--sync=0 \
$params_w \
--threads=1 \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/benchmark_filluniquerandom.log"
echo $cmd | tee $output_dir/benchmark_filluniquerandom.log
eval $cmd
summarize_result $output_dir/benchmark_filluniquerandom.log filluniquerandom filluniquerandom
}
function run_readrandom {
echo "Reading $num_keys random keys"
out_name="benchmark_readrandom.t${num_threads}.log"
cmd="./db_bench --benchmarks=readrandom \
--use_existing_db=1 \
$params_w \
--threads=$num_threads \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/${out_name}"
echo $cmd | tee $output_dir/${out_name}
eval $cmd
summarize_result $output_dir/${out_name} readrandom.t${num_threads} readrandom
}
function run_readwhile {
operation=$1
echo "Reading $num_keys random keys while $operation"
out_name="benchmark_readwhile${operation}.t${num_threads}.log"
cmd="./db_bench --benchmarks=readwhile${operation} \
--use_existing_db=1 \
--sync=$syncval \
$params_w \
--threads=$num_threads \
--merge_operator=\"put\" \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/${out_name}"
echo $cmd | tee $output_dir/${out_name}
eval $cmd
summarize_result $output_dir/${out_name} readwhile${operation}.t${num_threads} readwhile${operation}
}
function run_rangewhile {
operation=$1
full_name=$2
reverse_arg=$3
out_name="benchmark_${full_name}.t${num_threads}.log"
echo "Range scan $num_keys random keys while ${operation} for reverse_iter=${reverse_arg}"
cmd="./db_bench --benchmarks=seekrandomwhile${operation} \
--use_existing_db=1 \
--sync=$syncval \
$params_w \
--threads=$num_threads \
--merge_operator=\"put\" \
--seek_nexts=$num_nexts_per_seek \
--reverse_iterator=$reverse_arg \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/${out_name}"
echo $cmd | tee $output_dir/${out_name}
eval $cmd
summarize_result $output_dir/${out_name} ${full_name}.t${num_threads} seekrandomwhile${operation}
}
function run_range {
full_name=$1
reverse_arg=$2
out_name="benchmark_${full_name}.t${num_threads}.log"
echo "Range scan $num_keys random keys for reverse_iter=${reverse_arg}"
cmd="./db_bench --benchmarks=seekrandom \
--use_existing_db=1 \
$params_w \
--threads=$num_threads \
--seek_nexts=$num_nexts_per_seek \
--reverse_iterator=$reverse_arg \
--seed=$( date +%s ) \
2>&1 | tee -a $output_dir/${out_name}"
echo $cmd | tee $output_dir/${out_name}
eval $cmd
summarize_result $output_dir/${out_name} ${full_name}.t${num_threads} seekrandom
}
function run_randomtransaction {
echo "..."
cmd="./db_bench $params_r --benchmarks=randomtransaction \
--num=$num_keys \
--transaction_db \
--threads=5 \
--transaction_sets=5 \
2>&1 | tee $output_dir/benchmark_randomtransaction.log"
echo $cmd | tee $output_dir/benchmark_rangescanwhilewriting.log
eval $cmd
}
function now() {
echo `date +"%s"`
}
report="$output_dir/report.txt"
schedule="$output_dir/schedule.txt"
echo "===== Benchmark ====="
# Run!!!
IFS=',' read -a jobs <<< $1
for job in ${jobs[@]}; do
if [ $job != debug ]; then
echo "Start $job at `date`" | tee -a $schedule
fi
start=$(now)
if [ $job = bulkload ]; then
run_bulkload
elif [ $job = fillseq_disable_wal ]; then
run_fillseq 1
elif [ $job = fillseq_enable_wal ]; then
run_fillseq 0
elif [ $job = overwrite ]; then
run_change overwrite
elif [ $job = updaterandom ]; then
run_change updaterandom
elif [ $job = mergerandom ]; then
run_change mergerandom
elif [ $job = filluniquerandom ]; then
run_filluniquerandom
elif [ $job = readrandom ]; then
run_readrandom
elif [ $job = fwdrange ]; then
run_range $job false
elif [ $job = revrange ]; then
run_range $job true
elif [ $job = readwhilewriting ]; then
run_readwhile writing
elif [ $job = readwhilemerging ]; then
run_readwhile merging
elif [ $job = fwdrangewhilewriting ]; then
run_rangewhile writing $job false
elif [ $job = revrangewhilewriting ]; then
run_rangewhile writing $job true
elif [ $job = fwdrangewhilemerging ]; then
run_rangewhile merging $job false
elif [ $job = revrangewhilemerging ]; then
run_rangewhile merging $job true
elif [ $job = randomtransaction ]; then
run_randomtransaction
elif [ $job = universal_compaction ]; then
run_univ_compaction
elif [ $job = debug ]; then
num_keys=1000; # debug
echo "Setting num_keys to $num_keys"
else
echo "unknown job $job"
exit
fi
end=$(now)
if [ $job != debug ]; then
echo "Complete $job in $((end-start)) seconds" | tee -a $schedule
fi
echo -e "ops/sec\tmb/sec\tSize-GB\tL0_MB\tSum_GB\tW-Amp\tW-MB/s\tusec/op\tp50\tp75\tp99\tp99.9\tp99.99\tUptime\tStall-time\tStall%\tTest"
tail -1 $output_dir/report.txt
done