德鲁伊能否执行嵌套查询，使每个查询包含一个维度和一个关联维度列表？

Question

例如，给定此数据：

{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "1" "user": "abc" } 
{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "1" "user": "def" } 
{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "1" "user": "hgi" }
{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "2" "user": "mno" }
{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "2" "user": "qrs" } 
{"timestamp": "2011-01-12T00:00:00.000Z", "ip": "3" "user": "xyz" }

是否可以进行 returns

的高效查询

  {
    "timestamp": "...",
    "event": {
      "ip": 1,
      "user": ["abc", "def", "hgi"]
    },
  {
    "timestamp": "...",
    "event": {
      "ip": 2,
      "user": ["mno", "qrs"]
    },
  {
    "timestamp": "...",
    "event": {
      "ip": 3,
      "user": ["xyz"]
    }

如果是这样，是否可以仅限制 user 列表结果的结果计数？

Answer 1

对于德鲁伊 groupBy 查询，您不能应用 "sub-group" 或 "group_concat" 函数。这些根本不可用。 Druid 将根据您 select 的字段对您的查询进行分组。

当然，您可以按 ip 分组，然后计算行数甚至不同用户的数量。

德鲁伊能否执行嵌套查询，使每个查询包含一个维度和一个关联维度列表？

Can druid perform a nested query such that each contain one dimension and a list of associated dimensions?

druid