Question

SELECT ST.Id,
       ST.Label,
       ST.Asset,
       CASE
           WHEN ST.RFID = '' THEN 'NULL'
           ELSE ST.RFID
       END AS RFID,
       CASE
           WHEN ST.Type = 'O' THEN 'Odometer'
           ELSE 'Engine Hours'
       END AS AccumulatorType,
       ST.AppId,
       ST.Timestamp,
       ST.Accumulator,
       CASE
           WHEN ST.Flag = 0 THEN 'Disabled'
           WHEN ST.Flag = 1 THEN 'Active'
           WHEN ST.Flag = 3 THEN 'Rented'
           WHEN ST.Flag = 4 THEN 'Bypass'
           WHEN ST.Flag = 5 THEN 'Tanker'
       END AS TYPE,

  (SELECT COUNT(*)
   FROM sync
   WHERE RowId = ST.Id
     AND DefinitionId = 1
     AND Status = 1) AS Updated,

  (SELECT COUNT(*)
   FROM sync
   WHERE RowId = ST.Id
     AND DefinitionId = 1
     AND Status = 0) AS Remaining
FROM SecondaryTags AS ST
WHERE AppId = @AppId

Answer 1

为什么要重写逻辑？由于您在外部查询中进行了过滤，因此子查询可能是计算性能最高的方法。

可以通过确保具有写索引来加快子查询的速度。在这种情况下，您需要在sync(RowId, DefinitionId, StatusId)上建立索引：

create index idx_sync_rowid_definitionid_statusid
    on sync(RowId, DefinitionId, StatusId)

您可以将查询重写为：

select . . .,
       s.updated, s.remaining
from SecondaryTags st join
     (select rowid, sum(status = 1) as updated, sum(status = 0) as remaining
      from sync s
      where s.definitionId = 1
      group by s.rowid
     ) s
     on s.rowid = st.id
where st.AppId = @AppId;

不过，从性能角度来看，我认为索引是一个更好的主意。

Answer 2

{
    "something":  "this\nis\na\nsample\nfile"
}

您好，有人可以帮我用联接而不是子查询来编写此查询吗？

2 个答案: