influxdb速度测试的bash脚本speichern介绍

简介

这是一个针对InfluxDB进行速度测试的bash脚本speichern。使用该脚本可以测试InfluxDB在插入和查询数据时的性能表现，并提供相应的测试结果。

使用方法

1.下载speichern.sh脚本，放置于本地

2.修改脚本中的配置文件，使其满足自己的测试需求

3.执行脚本，即可得到测试结果

配置文件

配置文件位于脚本的最上方，由以下三个变量组成：

1.数据数量，默认值为10000

DATAPOINTS=10000

2.插入数据的时长，默认值为10秒

INSERT_DURATION_SEC=10

3.查询数据的时长，默认值为10秒

QUERY_DURATION_SEC=10

根据自己的需求修改对应的变量即可。

执行脚本

执行脚本的命令为：

bash speichern.sh

执行结果为JSON格式数据，包含了插入和查询测试的结果。

结果解读

结果中包含了InfluxDB在插入和查询数据时的吞吐量和延迟等信息，分别如下：

• Insertion Throughput - 插入吞吐量，以数据点每秒（Data points/second）为单位
• Insertion Latency - 插入延迟，以毫秒（ms）为单位
• Query Throughput - 查询吞吐量，以查询结果每秒（Query results/second）为单位
• Query Latency - 查询延迟，以毫秒（ms）为单位

这些信息可以帮助我们更好地评估InfluxDB在实际应用场景中的性能表现。

代码片段

#!/bin/bash

# Number of data points to write/read for each query
DATAPOINTS=10000

# Duration of data insertion period for each test
INSERT_DURATION_SEC=10

# Duration of data query period for each test
QUERY_DURATION_SEC=10

# Function to run a curl command and get the execution time from it
function time_curl {
    start=$(date -uSs)
    response=$(curl -XPOST "http://localhost:8086/write?db=test" --silent --data-binary "$1")
    end=$(date -uSs)

    elapsed=$(($(date -ud "$end" +%s%N) - $(date -ud "$start" +%s%N)))
    echo "$response" | grep "error" >/dev/null && echo "Error: $response"

    echo $elapsed
}

# Function to insert a specified number of data points into InfluxDB,
# measuring the throughput and latency of the writes
function run_insertion_test {
    echo "running $DATAPOINTS point insert test for $INSERT_DURATION_SEC seconds"
    data_points=""
    for i in $(seq 1 $DATAPOINTS); do
        data_points+="cpu_load_short,host=server01,region=us-west value=${i}.0\n"
    done

    elapsed_milliseconds=$(($(time_curl "$data_points") / 1000))
    insertion_throughput=$(echo "scale=2; $DATAPOINTS / $elapsed_milliseconds" | bc -l)
    insertion_latency=$(echo "scale=2; $elapsed_milliseconds / $DATAPOINTS" | bc -l)

    echo "insertion throughput: $insertion_throughput data points/second"
    echo "insertion latency: $insertion_latency ms/data point"
}

# Function to query for a specified number of data points from InfluxDB,
# measuring the throughput and latency of the read
function run_query_test {
    echo "running query test for $QUERY_DURATION_SEC seconds"
    time_to_stop=$(($(date -ud "$QUERY_DURATION_SEC seconds" +%s%N) / 1000000))

    queries=""
    until [ $(date +%s%N) -gt $time_to_stop ]; do
        query=$(cat <<-EOF
    SELECT mean(value) FROM cpu_load_short WHERE time >= $((DATAPOINTS / 2 + 1000000000))s+$((1000 * (${RANDOM}%4)))ms
    AND time <= $((DATAPOINTS / 2 + 1000000000))s+$((1000 * (${RANDOM}%4)))ms AND host='server01' AND region='us-west'
    EOF
        )

        queries+=$(echo -e "$query\n")
    done

    response=$(curl -XPOST "http://localhost:8086/query?db=test" --silent --data "$queries")
    query_results=$(echo "$response" | grep -c "\"name\":\"cpu_load_short\"")

    elapsed_milliseconds=$(($(date -uSs) / 1000 - START_TIME))
    query_throughput=$(echo "scale=2; $query_results / $elapsed_milliseconds" | bc -l)
    query_latency=$(echo "scale=2; $elapsed_milliseconds / $query_results" | bc -l)

    echo "query throughput: $query_throughput query results/second"
    echo "query latency: $query_latency ms/query result"
}

START_TIME=$(date -uSs) / 1000
run_insertion_test
run_query_test

# Return results in JSON format
echo "{
    \"insertion_throughput\": \"$insertion_throughput\",
    \"insertion_latency\": \"$insertion_latency\",
    \"query_throughput\": \"$query_throughput\",
    \"query_latency\": \"$query_latency\"
}" | jq

以上就是speichern脚本的介绍和相应的代码片段。