From b4e59a48fd204d1fceb6e2efd01d196ce7e8931f Mon Sep 17 00:00:00 2001 From: Levi Tamasi Date: Tue, 12 Oct 2021 11:33:36 -0700 Subject: [PATCH] Add a benchmarking wrapper script for BlobDB (#9015) Summary: The patch adds a new BlobDB benchmarking script called `run_blob_bench.sh`. It is a thin wrapper around `benchmark.sh` (similarly to `run_flash_bench.sh`): it actually calls `benchmark.sh` a number of times, cycling through six workloads, two write-only ones (bulk load and overwrite), two read/write ones (point lookups while writing, range scans while writing), and two read-only ones (point lookups and range scans). Note: this is a simpler/cleaned up/reworked version of the script used to produce the benchmark results in http://rocksdb.org/blog/2021/05/26/integrated-blob-db.html . The new version takes advantage of several recent `benchmark.sh` improvements like the ability to pass in arbitrary `db_bench` options or the possibility of using a job ID. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9015 Test Plan: Ran the script manually with different parameter combinations. Reviewed By: riversand963 Differential Revision: D31555277 Pulled By: ltamasi fbshipit-source-id: 0e151b2f7b2cf6f66ed7f95455571492ad7ea87f --- tools/benchmark.sh | 2 +- tools/run_blob_bench.sh | 195 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 196 insertions(+), 1 deletion(-) create mode 100755 tools/run_blob_bench.sh diff --git a/tools/benchmark.sh b/tools/benchmark.sh index 665990dfb..80e36f495 100755 --- a/tools/benchmark.sh +++ b/tools/benchmark.sh @@ -446,7 +446,7 @@ function run_fillseq { --disable_wal=$1 \ --seed=$( date +%s ) \ 2>&1 | tee -a $log_file_name" - + if [[ "$job_id" != "" ]]; then echo "Job ID: ${job_id}" > $log_file_name echo $cmd | tee -a $log_file_name diff --git a/tools/run_blob_bench.sh b/tools/run_blob_bench.sh new file mode 100755 index 000000000..b3657438a --- /dev/null +++ b/tools/run_blob_bench.sh @@ -0,0 +1,195 @@ +#!/usr/bin/env bash +# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved. +# +# BlobDB benchmark script +# +# REQUIRES: benchmark.sh is in the tools subdirectory +# +# After the execution of this script, log files are available in $output_dir. +# report.tsv provides high level statistics. +# +# Should be run from the parent of the tools directory. The command line is: +# [$env_vars] tools/run_blob_bench.sh +# +# This runs the following sequence of BlobDB performance tests: +# phase 1) write-only - bulkload+compact, overwrite+waitforcompaction +# phase 2) read-write - readwhilewriting, fwdrangewhilewriting +# phase 3) read-only - readrandom, fwdrange +# + +# Exit Codes +EXIT_INVALID_ARGS=1 + +# Size constants +K=1024 +M=$((1024 * K)) +G=$((1024 * M)) +T=$((1024 * G)) + +function display_usage() { + echo "usage: run_blob_bench.sh [--help]" + echo "" + echo "Runs the following sequence of BlobDB benchmark tests using tools/benchmark.sh:" + echo -e "\tPhase 1: write-only tests: bulkload+compact, overwrite+waitforcompaction" + echo -e "\tPhase 2: read-write tests: readwhilewriting, fwdrangewhilewriting" + echo -e "\tPhase 3: read-only tests: readrandom, fwdrange" + echo "" + echo "Environment Variables:" + echo -e "\tJOB_ID\t\t\t\tIdentifier for the benchmark job, will appear in the results (default: empty)" + echo -e "\tDB_DIR\t\t\t\tPath for the RocksDB data directory (mandatory)" + echo -e "\tWAL_DIR\t\t\t\tPath for the RocksDB WAL directory (mandatory)" + echo -e "\tOUTPUT_DIR\t\t\tPath for the benchmark results (mandatory)" + echo -e "\tNUM_THREADS\t\t\tNumber of threads (default: 16)" + echo -e "\tCOMPRESSION_TYPE\t\tCompression type for the SST files (default: lz4)" + echo -e "\tDB_SIZE\t\t\t\tRaw (uncompressed) database size (default: 1 TB)" + echo -e "\tVALUE_SIZE\t\t\tValue size (default: 1 KB)" + echo -e "\tNUM_KEYS\t\t\tNumber of keys (default: raw database size divided by value size)" + echo -e "\tDURATION\t\t\tIndividual duration for read-write/read-only tests in seconds (default: 1800)" + echo -e "\tWRITE_BUFFER_SIZE\t\tWrite buffer (memtable) size (default: 1 GB)" + echo -e "\tENABLE_BLOB_FILES\t\tEnable blob files (default: 1)" + echo -e "\tMIN_BLOB_SIZE\t\t\tSize threshold for storing values in blob files (default: 0)" + echo -e "\tBLOB_FILE_SIZE\t\t\tBlob file size (default: same as write buffer size)" + echo -e "\tBLOB_COMPRESSION_TYPE\t\tCompression type for the blob files (default: lz4)" + echo -e "\tENABLE_BLOB_GC\t\t\tEnable blob garbage collection (default: 1)" + echo -e "\tBLOB_GC_AGE_CUTOFF\t\tBlob garbage collection age cutoff (default: 0.25)" + echo -e "\tBLOB_GC_FORCE_THRESHOLD\t\tThreshold for forcing garbage collection of the oldest blob files (default: 1.0)" + echo -e "\tTARGET_FILE_SIZE_BASE\t\tTarget SST file size for compactions (default: write buffer size, scaled down if blob files are enabled)" + echo -e "\tMAX_BYTES_FOR_LEVEL_BASE\tMaximum size for the base level (default: 8 * target SST file size)" +} + +if [ $# -ge 1 ]; then + display_usage + + if [ "$1" == "--help" ]; then + exit + else + exit $EXIT_INVALID_ARGS + fi +fi + +# shellcheck disable=SC2153 +if [ -z "$DB_DIR" ]; then + echo "DB_DIR is not defined" + exit $EXIT_INVALID_ARGS +fi + +# shellcheck disable=SC2153 +if [ -z "$WAL_DIR" ]; then + echo "WAL_DIR is not defined" + exit $EXIT_INVALID_ARGS +fi + +# shellcheck disable=SC2153 +if [ -z "$OUTPUT_DIR" ]; then + echo "OUTPUT_DIR is not defined" + exit $EXIT_INVALID_ARGS +fi + +# shellcheck disable=SC2153 +job_id=$JOB_ID + +db_dir=$DB_DIR +wal_dir=$WAL_DIR +output_dir=$OUTPUT_DIR + +num_threads=${NUM_THREADS:-16} + +compression_type=${COMPRESSION_TYPE:-lz4} + +db_size=${DB_SIZE:-$((1 * T))} +value_size=${VALUE_SIZE:-$((1 * K))} +num_keys=${NUM_KEYS:-$((db_size / value_size))} + +duration=${DURATION:-1800} + +write_buffer_size=${WRITE_BUFFER_SIZE:-$((1 * G))} + +enable_blob_files=${ENABLE_BLOB_FILES:-1} +min_blob_size=${MIN_BLOB_SIZE:-0} +blob_file_size=${BLOB_FILE_SIZE:-$write_buffer_size} +blob_compression_type=${BLOB_COMPRESSION_TYPE:-lz4} +enable_blob_garbage_collection=${ENABLE_BLOB_GC:-1} +blob_garbage_collection_age_cutoff=${BLOB_GC_AGE_CUTOFF:-0.25} +blob_garbage_collection_force_threshold=${BLOB_GC_FORCE_THRESHOLD:-1.0} + +if [ "$enable_blob_files" == "1" ]; then + target_file_size_base=${TARGET_FILE_SIZE_BASE:-$((32 * write_buffer_size / value_size))} +else + target_file_size_base=${TARGET_FILE_SIZE_BASE:-$write_buffer_size} +fi + +max_bytes_for_level_base=${MAX_BYTES_FOR_LEVEL_BASE:-$((8 * target_file_size_base))} + +echo "======================== Benchmark setup ========================" +echo -e "Job ID:\t\t\t\t\t$job_id" +echo -e "Data directory:\t\t\t\t$db_dir" +echo -e "WAL directory:\t\t\t\t$wal_dir" +echo -e "Output directory:\t\t\t$output_dir" +echo -e "Number of threads:\t\t\t$num_threads" +echo -e "Compression type for SST files:\t\t$compression_type" +echo -e "Raw database size:\t\t\t$db_size" +echo -e "Value size:\t\t\t\t$value_size" +echo -e "Number of keys:\t\t\t\t$num_keys" +echo -e "Duration of read-write/read-only tests:\t$duration" +echo -e "Write buffer size:\t\t\t$write_buffer_size" +echo -e "Blob files enabled:\t\t\t$enable_blob_files" +echo -e "Blob size threshold:\t\t\t$min_blob_size" +echo -e "Blob file size:\t\t\t\t$blob_file_size" +echo -e "Compression type for blob files:\t$blob_compression_type" +echo -e "Blob GC enabled:\t\t\t$enable_blob_garbage_collection" +echo -e "Blob GC age cutoff:\t\t\t$blob_garbage_collection_age_cutoff" +echo -e "Blob GC force threshold:\t\t$blob_garbage_collection_force_threshold" +echo -e "Target SST file size:\t\t\t$target_file_size_base" +echo -e "Maximum size of base level:\t\t$max_bytes_for_level_base" +echo "=================================================================" + +rm -rf "$db_dir" +rm -rf "$wal_dir" +rm -rf "$output_dir" + +ENV_VARS="\ + JOB_ID=$job_id \ + DB_DIR=$db_dir \ + WAL_DIR=$wal_dir \ + OUTPUT_DIR=$output_dir \ + NUM_THREADS=$num_threads \ + COMPRESSION_TYPE=$compression_type \ + VALUE_SIZE=$value_size \ + NUM_KEYS=$num_keys" + +ENV_VARS_D="$ENV_VARS DURATION=$duration" + +PARAMS="\ + --enable_blob_files=$enable_blob_files \ + --min_blob_size=$min_blob_size \ + --blob_file_size=$blob_file_size \ + --blob_compression_type=$blob_compression_type \ + --write_buffer_size=$write_buffer_size \ + --target_file_size_base=$target_file_size_base \ + --max_bytes_for_level_base=$max_bytes_for_level_base" + +PARAMS_GC="$PARAMS \ + --enable_blob_garbage_collection=$enable_blob_garbage_collection \ + --blob_garbage_collection_age_cutoff=$blob_garbage_collection_age_cutoff \ + --blob_garbage_collection_force_threshold=$blob_garbage_collection_force_threshold" + +# bulk load (using fillrandom) + compact +env -u DURATION -S "$ENV_VARS" ./tools/benchmark.sh bulkload "$PARAMS" + +# overwrite + waitforcompaction +env -u DURATION -S "$ENV_VARS" ./tools/benchmark.sh overwrite "$PARAMS_GC" + +# readwhilewriting +env -S "$ENV_VARS_D" ./tools/benchmark.sh readwhilewriting "$PARAMS_GC" + +# fwdrangewhilewriting +env -S "$ENV_VARS_D" ./tools/benchmark.sh fwdrangewhilewriting "$PARAMS_GC" + +# readrandom +env -S "$ENV_VARS_D" ./tools/benchmark.sh readrandom "$PARAMS_GC" + +# fwdrange +env -S "$ENV_VARS_D" ./tools/benchmark.sh fwdrange "$PARAMS_GC" + +# save logs to output directory +cp "$db_dir"/LOG* "$output_dir/"