Running the Benchmark Test

Prerequisites

A fully operational Iguazio Data Science Platform setup (with data and application nodes)

Configuration

Create a Custom TSDB Configuration File

You can optionally create a TSDB configuration file instead of setting the relevant configuration using TSDB CLI configuration flags. Use the example configuration-file template at examples/ as a reference. By default, the TSDB CLI (tsbbctl) looks for a v3io-tsdb-config.yaml file in the current directory.

Create a TSDB Instance Using the TSDB CLI

Use the create command of the TSDB CLI to create a new TSDB instance (table). Run tsdbctl create -h for detailed information. You can use the following shell script as a reference:

Note: The script assumes that all configuration and executable files reside at the ${HOME}/go/bin directory.

    #!/bin/bash
    
    if [ "$1" == "" ]; then
      TSDB_PATH="pmetric"
    else
      TSDB_PATH=$1
    fi
    
    echo "Creating a TSDB instance at '$TSDB_PATH' ..."
    
    # Create a TSDB instance "tsdb" with a sampling rate of of one sample per
    # second, some aggregates, and an aggregation interval of 5 seconds.
    # Use -v to run in verbose mode. Use the -g flag to point to the
    # configuration file that you created in the previous step, or set the
    # web-gateway service endpoint and authentication credentials and the
    # parent data container using the CLI -s, -u, -p, and -c flags.
    SCRIPTPATH="$( cd "$(dirname "$0")" ; pwd -P )"
    $SCRIPTPATH/tsdbctl create -t $TSDB_PATH -r 1/s -a count,sum,min,max -i 5 -v -g $SCRIPTPATH/v3io-tsdb-custom.yaml
    
    echo Done.

Create a Benchmark-Test Ingestion Configuration File

Create a tsdb-bench-test-config.yaml configuration file for the TSDB ingestion benchmark test. Use test/benchmark/testdata/tsdb-bench-test-config.yaml as a reference.

Define Required Environment Variables

Define the following environment variables.

Note: You can also define variables locally in a script or in the command line from which the relevant command is executed.

    V3IO_TSDB_CONFIG="$HOME/go/bin/v3io-tsdb-custom.yaml"
    TSDB_BENCH_INGEST_CONFIG="$HOME/go/bin/tsdb-bench-test-config.yaml"

Run the Ingestion Benchmark Test

Run the ingestion benchmark test to populate the TSDB, using the desired time interval.

Use the following script as a reference:

Note: You can pass a "[0-9]+[mhd]" test-duration parameter to the script.

For example: `./ingest.sh 5m` will run 5 minutes.

> By default, the test will run for one minute.

    #!/bin/bash
    
    #File: ingest.sh
    
    if [ "$1" == "" ]; then
      BENCH_TIME="1m"
    else
      BENCH_TIME="$1"
    fi
    
    echo "Ingesting samples (bench time = $BENCH_TIME) ..."
    
    cd $HOME/go/src/github.com/v3io/v3io-tsdb/cmd/tsdbctl
    
    # Note: You can select either the "-bench=^BenchmarkIngest$" or
    # "-bench=^BenchmarkIngestWithNuclio$" test.
    time V3IO_TSDB_CONFIG="$HOME/go/bin/v3io-tsdb-custom.yaml" TSDB_BENCH_INGEST_CONFIG="$HOME/go/bin/tsdb-bench-test-config.yaml" go test -benchtime $BENCH_TIME -run=DO_NOT_RUN_TESTS -bench=^BenchmarkIngest$ ../../test/benchmark
    
    echo Done

Run a TSDB Query

Use the querey command of the TSDB CLI (tsdbctl) to query the test TSDB instance. Run tsdbctl query -h for detailed information.

The test produces Name_(A..Z)_(0..20) metrics. You can use the following shell script example to query counts for all types of metrics generated by the test.

Note: The following script fetches a count aggregate calculation for the last 72 hours with an aggregation interval of 5 minutes (which is similar to the TSDB aggregation granularity).

    #!/bin/bash
    
    #File: query.sh
    
    GOBIN=$HOME/go/bin
    LOOK_BACK_DURATION=72h
    TSDB_AGGREGATION_INTERVAL=5m
    V3IO_TSDB_CONFIG=$GOBIN/v3io-tsdb-custom.yaml
    
    for x in {A..Z}
    do
      for ((i = 0; i < 10; i++))
      do
        echo Querying metric Name_${x}_${i} ...
        $GOBIN/tsdbctl query Name_${x}_${i} -a count -l $LOOK_BACK_DURATION -i $TSDB_AGGREGATION_INTERVAL -g $V3IO_CONF
      done
    done
    
    echo Done

Example: Calculate a Total Samples Count

    #!/bin/bash
   
    # File: count-all.sh
     
    echo Querying the TSDB ...
    
    GOBIN=$HOME/go/bin
    COUNT="$($GOBIN/query.sh | grep -v "v=0" | grep 2018 | cut -d'=' -f 2 | awk '{s+=$1} END {print s}')"
    
    echo Total count of 2018 metric samples with value 0: $COUNT

# Packages

# README