如何加速使用 DISTINCT 的 SQL 查询

Question

据我了解，最好的解决方案似乎是创建一个 INDEX，但我不确定应该为哪些列创建索引。这是我第一次使用 SQL 索引。

如果我从此查询中删除 DISTINCT 调用，我将在一秒钟内获得超过 1000 个结果。然而，使用 DISTINCT 调用它 returns 在 10 秒内得到结果（显然没有重复项）。

如果有人有任何替代解决方案，我会洗耳恭听。

这是查询（第二个SELECT是调用DISTINCT函数的地方）：

SELECT 
    Sku,
    Name,
    ccp.Polygon,
    MarketAvailability,
    Coverage,
    Range  
FROM
    (SELECT DISTINCT
         dbo.CatalogEntry.CatalogEntryId as Id,
         dbo.CatalogEntry.Code as Sku,
         CoverageNode.Name as Coverage,
         RangeNode.Name as [Range],
         (SELECT CatalogContentProperty.LongString
          FROM CatalogContentProperty                         
          WHERE MetaFieldName = 'ItemChartName' 
            AND (CatalogContentProperty.LongString IS NOT NULL) 
            AND (CatalogContentProperty.ObjectId = dbo.CatalogEntry.CatalogEntryId)) AS [Name],
         (SELECT CatalogContentProperty.LongString
          FROM CatalogContentProperty                         
          WHERE MetaFieldName = 'MarketAvailabilityDetailsCollection' 
            AND (CatalogContentProperty.LongString IS NOT NULL) 
            AND (CatalogContentProperty.ObjectId = dbo.CatalogEntry.CatalogEntryId)) AS MarketAvailability
     FROM            
         dbo.CatalogEntry 
     INNER JOIN
         dbo.NodeEntryRelation ON dbo.CatalogEntry.CatalogEntryId = dbo.NodeEntryRelation.CatalogEntryId 
     INNER JOIN
         dbo.CatalogNode AS CoverageNode ON dbo.NodeEntryRelation.CatalogNodeId = CoverageNode.CatalogNodeId 
     INNER JOIN
         dbo.CatalogNode AS RangeNode ON CoverageNode.ParentNodeId = RangeNode.CatalogNodeId 
     INNER JOIN
         dbo.CatalogContentProperty ON dbo.CatalogEntry.CatalogEntryId = dbo.CatalogContentProperty.ObjectId 
     INNER JOIN
         dbo.CatalogNode AS ModelNode ON RangeNode.ParentNodeId = ModelNode.CatalogNodeId 
     INNER JOIN
         dbo.CatalogNode AS BrandNode ON ModelNode.ParentNodeId = BrandNode.CatalogNodeId 
     WHERE        
         (dbo.CatalogEntry.ClassTypeId = N'Variation') AND
         (dbo.CatalogContentProperty.MetaFieldName = N'ItemIsChart') AND
         RangeNode.Name != 'C-MAP'  AND
         (BrandNode.Name = '' OR '' = '' OR '' IS NULL) AND
         (ModelNode.Name = '' OR '' = '' OR '' IS NULL) AND 
         (CoverageNode.Name = '' OR '' = '' OR '' IS NULL) AND
         (RangeNode.Name = '' OR '' = '' OR '' IS NULL) 
    ) AS CmapResults 
INNER JOIN
    (SELECT   
         GEOMETRY::STGeomFromText(CatalogContentProperty.LongString,4326) AS PolygonGeometry,
         CatalogContentProperty.LongString AS Polygon,
         CatalogContentProperty.ObjectId
     FROM       
         CatalogContentProperty                         
     WHERE   
         MetaFieldName = 'ItemChartCoordinates' AND 
         (CatalogContentProperty.LongString IS NOT NULL)) ccp ON ccp.ObjectId = CmapResults.Id 
WHERE 
    ((GEOGRAPHY::STGeomFromText(PolygonGeometry.MakeValid().STUnion(PolygonGeometry.MakeValid().STStartPoint()).STAsText(), 4326).STDistance(GEOGRAPHY::STGeomFromText('POINT(50.9835929 -1.4205852)', 4326)) / 1609.344) <= 100 OR 'POINT(50.9835929 -1.4205852)' IS NULL)
  AND MarketAvailability IS NOT NULL
ORDER BY 
     GEOGRAPHY::STGeomFromText(PolygonGeometry.MakeValid().STUnion(PolygonGeometry.MakeValid().STStartPoint()).STAsText(), 4326).STArea() DESC;

我正在使用 SQL Server Management Studio 2012。目的是通过对 return 的 DISTINCT 调用在与没有 DISTINCT 调用的查询相同的时间内获取查询。

Answer 1

您的查询看起来有点复杂。这个会产生相同的输出吗？

SELECT DISTINCT ce.Code as Sku
, max(case when ccp.MetaFieldName = 'ItemChartName' then ccp.LongString end) as 'Name'
, max(case when ccp.MetaFieldName = 'MarketAvailabilityDetailsCollection' then ccp.LongString end) as 'MarketAvailability'
, max(case when ccp.MetaFieldName = 'ItemChartCoordinates' then ccp.LongString end) as 'Polygon'
, CoverageNode.Name as Coverage
, RangeNode.Name as 'Range'

FROM dbo.CatalogEntry ce
  INNER JOIN dbo.NodeEntryRelation ner ON ce.CatalogEntryId = ner.CatalogEntryId
  INNER JOIN dbo.CatalogNode CoverageNode ON dbo.NodeEntryRelation.CatalogNodeId = CoverageNode.CatalogNodeId
  INNER JOIN dbo.CatalogNode RangeNode ON CoverageNode.ParentNodeId = RangeNode.CatalogNodeId
  INNER JOIN dbo.CatalogContentProperty ccp ON ce.CatalogEntryId = ccp.ObjectId
  INNER JOIN dbo.CatalogNode ModelNode ON RangeNode.ParentNodeId = ModelNode.CatalogNodeId
  INNER JOIN dbo.CatalogNode BrandNode ON ModelNode.ParentNodeId = BrandNode.CatalogNodeId

WHERE ce.ClassTypeId = N'Variation'
  and ccp.MetaFieldName = N'ItemIsChart'
  and RangeNode.Name != 'C-MAP'
  --and (BrandNode.Name = '' OR '' = '' OR '' IS NULL)      --  always true
  --and (ModelNode.Name = '' OR '' = '' OR '' IS NULL)      --  always true
  --and (CoverageNode.Name = '' OR '' = '' OR '' IS NULL)   --  always true
  --and (RangeNode.Name = '' OR '' = '' OR '' IS NULL)      --  always true
  and (GEOGRAPHY::STGeomFromText(GEOMETRY::STGeomFromText(ccp.LongString,4326).MakeValid().STUnion(GEOMETRY::STGeomFromText(ccp.LongString,4326).MakeValid().STStartPoint()).STAsText(), 4326).STDistance(GEOGRAPHY::STGeomFromText('POINT(50.9835929 -1.4205852)', 4326)) / 1609.344) <= 100   -- OR 'POINT(50.9835929 -1.4205852)' IS NULL    --  redundant
)
and max(case when ccp.MetaFieldName = 'MarketAvailabilityDetailsCollection' then ccp.LongString end)

ORDER BY GEOGRAPHY::STGeomFromText(GEOMETRY::STGeomFromText(ccp.LongString,4326).MakeValid().STUnion(GEOMETRY::STGeomFromText(ccp.LongString,4326).MakeValid().STStartPoint()).STAsText(), 4326).STArea() DESC;

Answer 2

所以答案是我的情况是删除 DISTINCT 并使用 LINQ 进行重复删除！加载时间从 10-11 秒减少到 4-5 秒

如何加速使用 DISTINCT 的 SQL 查询

How to speed up a SQL query which is using DISTINCT

sql

indexing

distinct

sql-server-2012