1. 程式人生 > >通過 elasticsearch-sql 使用 SQL 語句聚合查詢 Elasticsearch 獲取各種 metrics 度量值

通過 elasticsearch-sql 使用 SQL 語句聚合查詢 Elasticsearch 獲取各種 metrics 度量值

Elasticsearch 的 metrics(度量)包含 count、sum、avg、max、min、percentiles (百分位數)、Unique count(基數 || 去重計數)、Median(中位數)、擴充套件度量(含方差、平方和、標準差、標準差界限)、Percentile ranks(百分位等級)

1.count(數量):

SELECT count(log_date.d) AS Count FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "COUNT" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "Count" : {
      "value_count" : {
        "field" : "log_date.d"
      }
    }
  }
}

2.sum(和)

SELECT sum(log_date.d) AS SUM FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "SUM" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "SUM" : {
      "sum" : {
        "field" : "log_date.d"
      }
    }
  }
}

3.avg(平均數)

SELECT avg(log_date.d) AS AVG FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "AVG" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "AVG" : {
      "avg" : {
        "field" : "log_date.d"
      }
    }
  }
}

4.max(最大值)

SELECT max(log_date.d) AS MAX FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "MAX" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "MAX" : {
      "max" : {
        "field" : "log_date.d"
      }
    }
  }
}

5.min(最小值)

SELECT min(log_date.d) AS MIN FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "MIN" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "MIN" : {
      "min" : {
        "field" : "log_date.d"
      }
    }
  }
}

6.percentiles(百分位數):

SELECT percentiles(log_date.d,1.0,15.0,31.0) AS Percentiles FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "percentiles" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "Percentiles" : {
      "percentiles" : {
        "field" : "log_date.d",
        "percents" : [ 1.0, 15.0, 31.0 ]
      }
    }
  }
}

7.Unique count(基數 || 去重計數,就是 SQL 中的 distinct ):

SELECT count(distinct(log_date.d)) AS UniqueCount FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "COUNT" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "UniqueCount" : {
      "cardinality" : {
        "field" : "log_date.d",
        "precision_threshold" : 40000
      }
    }
  }
}

8.Median(中位數):

中位數沒找到單獨的獲取方法,不過在 Kibana 中看到獲取中位數時請求中的引數,其實就是獲取的某個欄位50的百分位數,所以可能有:中位數=50的百分位數

SELECT percentiles(log_date.d,50.0) AS percentiles FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "percentiles" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "percentiles" : {
      "percentiles" : {
        "field" : "log_date.d",
        "percents" : [ 50.0 ]
      }
    }
  }
}

9.方差、平方和、標準差、標準差界限

這幾個度量沒有單獨方法去獲取,都是用 EXTENDED_STATS 一個請求全部獲取下來,然後從中取自己需要的結果

SELECT EXTENDED_STATS(log_date.d) AS EXTENDED_STATS FROM INDEX-2017-12
{
  "from" : 0,
  "size" : 0,
  "_source" : {
    "includes" : [ "EXTENDED_STATS" ],
    "excludes" : [ ]
  },
  "aggregations" : {
    "EXTENDED_STATS" : {
      "extended_stats" : {
        "field" : "log_date.d"
      }
    }
  }
}

EXTENDED_STATS 查詢結果包含:方差、平方和、標準差、標準差界限以及最大值、平均數等基礎度量,具體如下:

"aggregations": {
    "1": {
      "count": 15304326,
      "min": 1,
      "max": 31,
      "avg": 15.068216202399244,
      "sum": 230608893,
      "sum_of_squares": 4588588661,
      "variance": 72.7718426201877,
      "std_deviation": 8.530641395591992,
      "std_deviation_bounds": {
        "upper": 32.129498993583226,
        "lower": -1.9930665887847407
      }
    }
  }

10.Percentile ranks(百分位等級)

暫時沒找到求百分位等級的 SQL 語句,只能用原生 ES 查詢語句獲取了;

ES原生查詢語句如下:

{
  "size": 0,
  ......
  "aggs": {
    "1": {
      "percentile_ranks": {
        "field": "log_date.d",
        "values": [
          6,
          15,
          31
        ],
        "keyed": false
      }
    }
  }
}