Question

数据：

+-------+------------+-----------------------+
| value |   metric   |  timestamp            |
+-------+------------+-----------------------+
| 30    | tempA      |  2019-01-12T01:00:00  |
| 28    | tempA      |  2019-02-12T02:00:00  |
| 25    | tempB      |  2019-03-12T03:00:00  |
| 60    | humidityA  |  2019-04-12T04:00:00  |
| 15    | tempC      |  2019-05-15T01:00:00  |
+-------+------------+-----------------------+

我想获取某些温度指标的最新已知值。我想出了：

SELECT
  metric,
  value,
  timestamp,
  seqnum
FROM (
  SELECT
    metric,
    value,
    timestamp,
    ROW_NUMBER() OVER (PARTITION BY metric ORDER BY timestamp DESC) AS seqnum
  FROM
    `project.dataset.table`
  WHERE
    metric IN ('tempA', 'tempB') )
WHERE
  seqnum = 1

这会对整个表进行分区，这可能需要一些时间。是否有更高效/高效的方法来获取每个指标的最后一行？

Answer 1

数组聚合可以更快，因为由于LIMIT 1，它只需要将第一行保留在内存中即可。

SELECT
  metric,
  ARRAY_AGG(
    STRUCT(value, timestamp)
    ORDER BY timestamp DESC LIMIT 1
  )[OFFSET(0)].*
FROM
  `project.dataset.table`
WHERE
  metric IN ('tempA', 'tempB')
GROUP BY metric

如何获取每个指标BigQuery的最后一行？

1 个答案: