Question

我在c＃中使用以下sqlite查询。查询将针对100K项进行迭代。我觉得执行速度很慢，通过优化查询可以提高速度

insert into salesmetrics(salesid, salesrankcount, volumerankcount, countsales,Avgsales) 
select @salesid,sum(salesrank), sum(volumerank), Count(salesrank), avg(salesrank) 
from (select salesrank, volumerank
      from salesindex
      Join SalesData on salesindex.salesindexID = salesData.salesindexID
      where SourceID = @Sourceid
        and Content like @content
      group by salesindex.salesindexID)

表：

CREATE TABLE `salesindex` (
    salesindexID INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT,
    salesindex TEXT,
    SourceID INTEGER,
    salesrank INTEGER,
    volumerank INTEGER,
    dateAdded DATETIME, 
    UNIQUE(SourceID,dateAdded)
);

CREATE TABLE `SalesData` (
    SalesDataID INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT,
    salesindexID INTEGER,
    Content TEXT,
    dateAdded DATETIME,
    UNIQUE(salesindexID,content)
);

CREATE TABLE `salesmetrics` (
   salesmetricsID INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT,
   salesID INTEGER,
   SalesRankCount INTEGER,
   VolumeRankCount INTEGER,
   countsales INTEGER,
   SalesRank  FLOAT
);

salesdata表上的“content”和salesindex表上的sourceid的索引

Answer 1

主要问题是内部查询中的GROUP BY，显然需要压缩多个匹配的SalesData行。

执行Content检查子查询：

SELECT @salesid,
       SUM(salesrank),
       SUM(volumerank),
       COUNT(salesrank),
       AVG(salesrank)
FROM salesindex
WHERE salesindexID IN (SELECT salesindexID
                       FROM SalesData
                       WHERE Content LIKE @content)
  AND SourceID = @Sourceid;

或者，根据Content查找的选择性，将查找作为相关子查询进行查询可能更好：

SELECT @salesid,
       SUM(salesrank),
       SUM(volumerank),
       COUNT(salesrank),
       AVG(salesrank)
FROM salesindex
WHERE EXISTS (SELECT 1
              FROM SalesData
              WHERE salesindexID = salesindex.salesindexID
                AND Content LIKE @content)
  AND SourceID = @Sourceid;

Answer 2

除非我遗漏了某些内容，否则我认为您可以直接插入选择，而不是选择选择：

insert into salesmetrics(salesid, salesrankcount, volumerankcount, countsales,Avgsales)
(select @salesid,sum(salesrank), sum(volumerank), Count(salesrank), avg(salesrank)
 from salesindex 
 Join SalesData on salesindex.salesindexID = salesData.salesindexID
 where SourceID = @Sourceid
   and Content like @content
 group by salesindex.salesindexID)

使用嵌套查询的sqlite查询性能

2 个答案: