From 8dccb6a125885d9489c95ce950ce6fa173a2d69a Mon Sep 17 00:00:00 2001 From: Flook Peter Date: Thu, 4 Jul 2024 14:27:13 +0800 Subject: [PATCH] Add in comparison script to check for performance differences with previous version --- .github/workflows/benchmark.yml | 5 +- benchmark/compare_benchmark_results.sh | 48 +++++++++++++++ .../results/benchmark_results_0.10.6.txt | 60 +++++++++---------- .../results/benchmark_results_0.10.7.txt | 60 +++++++++---------- .../results/benchmark_results_0.10.9.txt | 60 +++++++++---------- .../results/benchmark_results_0.11.2.txt | 60 +++++++++---------- .../results/benchmark_results_0.11.5.txt | 60 +++++++++---------- benchmark/results/benchmark_results_0.4.5.txt | 60 +++++++++---------- benchmark/results/benchmark_results_0.6.0.txt | 60 +++++++++---------- benchmark/run_benchmark.sh | 5 +- 10 files changed, 263 insertions(+), 215 deletions(-) create mode 100644 benchmark/compare_benchmark_results.sh diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml index 32aff04..ae6cd27 100644 --- a/.github/workflows/benchmark.yml +++ b/.github/workflows/benchmark.yml @@ -32,7 +32,10 @@ jobs: - name: Get Spark query engine jars run: bash benchmark/setup_query_engine_jars.sh - name: Run benchmark script - run: bash benchmark/run_benchmark.sh + run: | + version=$(grep dataCatererVersion gradle.properties | cut -d= -f2) + bash benchmark/run_benchmark.sh + bash benchmark/compare_benchmark_results.sh "$version" - name: Create pull request uses: peter-evans/create-pull-request@v6 with: diff --git a/benchmark/compare_benchmark_results.sh b/benchmark/compare_benchmark_results.sh new file mode 100644 index 0000000..7c9584f --- /dev/null +++ b/benchmark/compare_benchmark_results.sh @@ -0,0 +1,48 @@ +#!/usr/bin/env bash + +SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" &>/dev/null && pwd) +LATEST_VERSION=${1:-0.11.5} +RESULT_FILE_REGEX="benchmark_results_([0-9\.]+)\.txt" + +if [[ -z $2 ]]; then + echo "No second version to compare against passed into arguments, defaulting to previous version in results" + PREVIOUS_VERSION_RESULT_FILE_NAME=$(ls -1 "$SCRIPT_DIR/results" | sort --version-sort | tail -2 | head -1) + if [[ $PREVIOUS_VERSION_RESULT_FILE_NAME =~ $RESULT_FILE_REGEX ]]; then + PREVIOUS_VERSION="${BASH_REMATCH[1]}" + else + echo "Previous version file name does not match regex: $RESULT_FILE_REGEX, previous version file: $PREVIOUS_VERSION_RESULT_FILE_NAME" + exit 1 + fi +else + PREVIOUS_VERSION=${2} +fi +echo "Latest version: $LATEST_VERSION" +echo "Previous version: $PREVIOUS_VERSION" +echo + + +plans=( + "io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000," + "io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000," + "io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000," + "io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000," + "io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000," +) + +for plan in "${plans[@]}"; do + plan_name=$(echo "$plan" | sed 's/io.github.datacatering.plan.benchmark.//') + echo "Comparing performance for plan: $plan_name" + + latest_version_results=$(cat "$SCRIPT_DIR/results/benchmark_results_${LATEST_VERSION}.txt" | grep "${plan}") + previous_version_results=$(cat "$SCRIPT_DIR/results/benchmark_results_${PREVIOUS_VERSION}.txt" | grep "${plan}") + + latest_version_average_time=$(echo "$latest_version_results" | awk -F "," '{s+=$4} END {print s/3}') + previous_version_average_time=$(echo "$previous_version_results" | awk -F "," '{s+=$4} END {print s/3}') + + difference=$(awk -v t1="$previous_version_average_time" -v t2="$latest_version_average_time" 'BEGIN{printf "%.3f", t2-t1}') + percent_difference=$(awk -v t1="$previous_version_average_time" -v t2="$latest_version_average_time" 'BEGIN{printf "%.3f", (t2-t1)/t1 * 100}') + + echo "Version: $PREVIOUS_VERSION, Average time (s): $previous_version_average_time" + echo "Version: $LATEST_VERSION, Average time (s): $latest_version_average_time, Difference (s): $difference, Percent: $percent_difference%" + echo +done diff --git a/benchmark/results/benchmark_results_0.10.6.txt b/benchmark/results/benchmark_results_0.10.6.txt index 7220808..52d8bae 100644 --- a/benchmark/results/benchmark_results_0.10.6.txt +++ b/benchmark/results/benchmark_results_0.10.6.txt @@ -8,33 +8,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 1, 18.10 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 2, 17.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 3, 18.75 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 1, 18.45 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 2, 18.55 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 3, 18.54 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 1, 18.97 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 2, 18.61 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 3, 18.74 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 1, 3.22 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 2, 3.11 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 3, 3.29 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 1, 13.15 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 2, 13.01 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 3, 13.17 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.55 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 18.76 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.72 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 1, 65.31 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 2, 66.68 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 3, 65.88 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 1, 37.88 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 2, 37.10 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 3, 38.23 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 1, 18.32 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 2, 18.64 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 3, 18.17 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.36 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 18.39 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.85 \ No newline at end of file +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,1,18.10 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,2,17.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,3,18.75 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,1,18.45 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,2,18.55 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,3,18.54 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,1,18.97 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,2,18.61 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,3,18.74 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,1,3.22 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,2,3.11 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,3,3.29 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,1,13.15 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,2,13.01 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,3,13.17 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.55 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,18.76 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.72 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,1,65.31 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,2,66.68 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,3,65.88 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,1,37.88 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,2,37.10 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,3,38.23 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,1,18.32 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,2,18.64 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,3,18.17 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.36 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,18.39 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.85 \ No newline at end of file diff --git a/benchmark/results/benchmark_results_0.10.7.txt b/benchmark/results/benchmark_results_0.10.7.txt index 1eafdd7..262e4ca 100644 --- a/benchmark/results/benchmark_results_0.10.7.txt +++ b/benchmark/results/benchmark_results_0.10.7.txt @@ -9,33 +9,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 1, 19.26 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 2, 19.02 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 3, 18.47 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 1, 19.11 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 2, 18.66 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 3, 19.17 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 1, 18.96 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 2, 18.58 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 3, 18.41 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 1, 3.31 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 2, 3.31 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 3, 3.27 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 1, 13.47 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 2, 13.51 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 3, 13.52 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.91 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 19.11 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.93 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 1, 67.12 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 2, 67.25 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 3, 67.23 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 1, 37.41 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 2, 38.44 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 3, 38.26 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 1, 19.30 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 2, 18.30 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 3, 18.88 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 18.42 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.96 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,1,19.26 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,2,19.02 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,3,18.47 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,1,19.11 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,2,18.66 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,3,19.17 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,1,18.96 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,2,18.58 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,3,18.41 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,1,3.31 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,2,3.31 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,3,3.27 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,1,13.47 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,2,13.51 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,3,13.52 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.91 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,19.11 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.93 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,1,67.12 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,2,67.25 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,3,67.23 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,1,37.41 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,2,38.44 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,3,38.26 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,1,19.30 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,2,18.30 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,3,18.88 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,18.42 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.96 diff --git a/benchmark/results/benchmark_results_0.10.9.txt b/benchmark/results/benchmark_results_0.10.9.txt index 25ff744..25dc4a3 100644 --- a/benchmark/results/benchmark_results_0.10.9.txt +++ b/benchmark/results/benchmark_results_0.10.9.txt @@ -9,33 +9,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 1, 18.62 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 2, 18.50 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 3, 18.22 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 1, 19.03 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 2, 18.51 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 3, 18.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 1, 18.43 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 2, 19.14 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 3, 18.70 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 1, 3.37 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 2, 3.20 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 3, 3.30 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 1, 13.30 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 2, 12.91 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 3, 13.26 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.33 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 18.41 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.49 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 1, 66.02 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 2, 63.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 3, 66.96 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 1, 37.58 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 2, 37.86 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 3, 37.33 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 1, 18.70 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 2, 18.41 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 3, 18.39 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 18.66 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 18.57 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 18.35 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,1,18.62 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,2,18.50 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,3,18.22 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,1,19.03 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,2,18.51 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,3,18.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,1,18.43 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,2,19.14 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,3,18.70 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,1,3.37 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,2,3.20 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,3,3.30 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,1,13.30 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,2,12.91 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,3,13.26 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.33 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,18.41 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.49 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,1,66.02 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,2,63.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,3,66.96 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,1,37.58 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,2,37.86 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,3,37.33 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,1,18.70 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,2,18.41 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,3,18.39 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,18.66 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,18.57 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,18.35 diff --git a/benchmark/results/benchmark_results_0.11.2.txt b/benchmark/results/benchmark_results_0.11.2.txt index 380c42a..35ab5e0 100644 --- a/benchmark/results/benchmark_results_0.11.2.txt +++ b/benchmark/results/benchmark_results_0.11.2.txt @@ -9,33 +9,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 1, 15.27 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 2, 15.05 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 3, 15.53 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 1, 15.28 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 2, 15.37 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 3, 15.45 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 1, 14.99 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 2, 14.86 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 3, 15.36 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 1, 15.19 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 2, 14.89 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 3, 15.24 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 1, 10.28 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 2, 10.07 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 3, 10.70 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 15.29 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 15.16 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 14.93 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 1, 50.52 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 2, 50.28 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 3, 50.12 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 1, 41.74 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 2, 41.70 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 3, 42.05 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 1, 15.34 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 2, 15.04 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 3, 15.47 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 15.06 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 15.11 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 15.19 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,1,15.27 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,2,15.05 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,3,15.53 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,1,15.28 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,2,15.37 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,3,15.45 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,1,14.99 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,2,14.86 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,3,15.36 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,1,15.19 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,2,14.89 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,3,15.24 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,1,10.28 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,2,10.07 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,3,10.70 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,15.29 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,15.16 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,14.93 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,1,50.52 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,2,50.28 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,3,50.12 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,1,41.74 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,2,41.70 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,3,42.05 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,1,15.34 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,2,15.04 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,3,15.47 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,15.06 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,15.11 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,15.19 diff --git a/benchmark/results/benchmark_results_0.11.5.txt b/benchmark/results/benchmark_results_0.11.5.txt index 54298a8..dc0828f 100644 --- a/benchmark/results/benchmark_results_0.11.5.txt +++ b/benchmark/results/benchmark_results_0.11.5.txt @@ -9,33 +9,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 1, 15.13 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 2, 15.39 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default, 100000, 3, 15.05 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 1, 14.39 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 2, 14.89 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze, 100000, 3, 15.29 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 1, 15.05 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 2, 15.53 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet, 100000, 3, 14.54 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 1, 15.07 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 2, 14.37 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten, 100000, 3, 14.60 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 1, 10.23 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 2, 10.33 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 10000, 3, 10.07 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 14.87 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 14.85 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 15.12 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 1, 51.25 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 2, 50.28 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 1000000, 3, 52.16 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 1, 41.75 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 2, 41.38 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:, 500000, 3, 41.08 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 1, 14.90 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 2, 14.85 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:, 100000, 3, 14.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 1, 15.22 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 2, 15.35 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:, 100000, 3, 14.96 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,1,15.13 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,2,15.39 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:default,100000,3,15.05 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,1,14.39 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,2,14.89 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:blaze,100000,3,15.29 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,1,15.05 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,2,15.53 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:comet,100000,3,14.54 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,1,15.07 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,2,14.37 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:gluten,100000,3,14.60 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,1,10.23 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,2,10.33 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,10000,3,10.07 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,14.87 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,14.85 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,15.12 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,1,51.25 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,2,50.28 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,1000000,3,52.16 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,1,41.75 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,2,41.38 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun:,500000,3,41.08 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,1,14.90 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,2,14.85 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun:,100000,3,14.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,1,15.22 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,2,15.35 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun:,100000,3,14.96 diff --git a/benchmark/results/benchmark_results_0.4.5.txt b/benchmark/results/benchmark_results_0.4.5.txt index 224fa9b..a3c080d 100644 --- a/benchmark/results/benchmark_results_0.4.5.txt +++ b/benchmark/results/benchmark_results_0.4.5.txt @@ -7,33 +7,33 @@ System info: CPUs: 4 Total Memory: 9.718GiB Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 1, 8.72 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 2, 8.62 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 3, 9.05 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 4, 8.30 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 5, 8.34 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 1, 12.19 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 2, 12.53 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 3, 12.25 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 4, 12.09 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 5, 12.28 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 1, 50.90 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 2, 53.87 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 3, 50.22 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 4, 49.40 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 5, 48.97 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 1, 26.06 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 2, 26.77 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 3, 26.65 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 4, 26.40 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 5, 26.79 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 1, 12.08 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 2, 12.14 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 3, 12.06 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 4, 12.23 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 5, 11.91 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 1, 12.31 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 2, 12.13 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 3, 12.08 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 4, 12.37 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 5, 12.18 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,1,8.72 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,2,8.62 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,3,9.05 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,4,8.30 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,5,8.34 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,1,12.19 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,2,12.53 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,3,12.25 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,4,12.09 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,5,12.28 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,1,50.90 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,2,53.87 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,3,50.22 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,4,49.40 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,5,48.97 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,1,26.06 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,2,26.77 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,3,26.65 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,4,26.40 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,5,26.79 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,1,12.08 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,2,12.14 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,3,12.06 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,4,12.23 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,5,11.91 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,1,12.31 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,2,12.13 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,3,12.08 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,4,12.37 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,5,12.18 diff --git a/benchmark/results/benchmark_results_0.6.0.txt b/benchmark/results/benchmark_results_0.6.0.txt index 4cc4de5..c85004e 100644 --- a/benchmark/results/benchmark_results_0.6.0.txt +++ b/benchmark/results/benchmark_results_0.6.0.txt @@ -9,33 +9,33 @@ System info: Driver memory: DRIVER_MEMORY=2g Executor memory: EXECUTOR_MEMORY=2g Class name, Num records, Num run, Time taken (s) -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 1, 9.52 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 2, 10.73 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 3, 9.18 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 4, 8.92 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 10000, 5, 8.53 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 1, 14.84 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 2, 14.09 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 3, 16.09 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 4, 12.95 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 5, 12.98 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 1, 60.89 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 2, 63.11 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 3, 59.45 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 4, 63.00 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 1000000, 5, 60.59 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 1, 28.32 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 2, 28.12 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 3, 27.04 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 4, 27.38 -io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun, 500000, 5, 27.51 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 1, 13.15 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 2, 14.21 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 3, 13.18 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 4, 13.08 -io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun, 100000, 5, 12.82 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 1, 12.74 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 2, 12.87 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 3, 12.82 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 4, 12.80 -io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun, 100000, 5, 12.70 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,1,9.52 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,2,10.73 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,3,9.18 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,4,8.92 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,10000,5,8.53 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,1,14.84 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,2,14.09 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,3,16.09 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,4,12.95 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,5,12.98 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,1,60.89 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,2,63.11 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,3,59.45 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,4,63.00 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,1000000,5,60.59 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,1,28.32 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,2,28.12 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,3,27.04 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,4,27.38 +io.github.datacatering.plan.benchmark.BenchmarkForeignKeyPlanRun,500000,5,27.51 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,1,13.15 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,2,14.21 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,3,13.18 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,4,13.08 +io.github.datacatering.plan.benchmark.BenchmarkJsonPlanRun,100000,5,12.82 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,1,12.74 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,2,12.87 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,3,12.82 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,4,12.80 +io.github.datacatering.plan.benchmark.BenchmarkParquetPlanRun,100000,5,12.70 diff --git a/benchmark/run_benchmark.sh b/benchmark/run_benchmark.sh index 2e8b23a..1c403b1 100755 --- a/benchmark/run_benchmark.sh +++ b/benchmark/run_benchmark.sh @@ -69,7 +69,7 @@ run_docker() { else final_record_count=$2 fi - echo "$1:$3, $final_record_count, $num_run, $time_taken" >> "$benchmark_result_file" + echo "$1:$3,$final_record_count,$num_run,$time_taken" >> "$benchmark_result_file" done } @@ -117,9 +117,6 @@ docker ps -a | grep -v "Exited (0)" | awk -F " " '{print $1}' | tail -1 | xargs echo "Printing logs of last docker run" docker ps -a | awk -F " " '{print $1}' | tail -1 | xargs docker logs -echo "Printing benchmark results" -cat "$benchmark_result_file" - echo "Cleaning docker runs..." docker ps -a | grep "datacatering/data-caterer-basic" | awk -F " " '{print $1}' | xargs docker rm echo "Done!"