如何加速复杂的MYSQL查询？

Question

我从ColdFusion网站调用Cold Fusion组件中有以下MYSQL查询。 我有一个自定义的，基本上为以下字段构建一个布尔查询：“PlaceName，Country，Adm1，adm2，adm3，locality”。

<!--- SEARCH PLACES 2--->
<cffunction name="searchPlaces_full" access="public" returntype="query">
    <cfargument name="q" type="string" required="yes">


    <CF_BOOLSEARCH searchterm="#arguments.q#" field="PlaceName,Country,Adm1,adm2,adm3,locality " booloperator="and">
    <cfquery name="GetPlaces" datasource="#application.settings.dsn#">
    SELECT 
        places.CatID,
        places.PlaceName,
        places.PlaceID,
        places.Address1,
        places.PostalCode,
        places.Locality,
        places.Address2,
        places.ImageThumb,
        places.Adm1,
        places.Country,
        places.Adm2,
        places.Adm3,
        places.contributorid,
        places.lng,
        places.lat,
        places.verified,
        places.verified_by,
        places.verified_date
    FROM places INNER JOIN places_cats ON places.PlaceID = places_cats.PlaceID
    WHERE  
    <cfif len(trim(arguments.q))>
    (#PreserveSingleQuotes(boolsearch)#)
    <cfelse>
    1=0
    </cfif>
    AND places_cats.CATID IN (#arguments.categories#)
    GROUP BY 
        places.CatID,
        places.PlaceName,
        places.PlaceID,
        places.Address1,
        places.PostalCode,
        places.Locality,
        places.Address2,
        places.ImageThumb,
        places.Adm1,
        places.Country,
        places.Adm2,
        places.Adm3,
        places.contributorid,
        places.lng,
        places.lat,
        places.verified,
        places.verified_by,
        places.verified_date
        ORDER BY PlaceName
        </cfquery>


    <cfreturn getPlaces>
</cffunction>

数据库中有624227条记录。 如果我搜索Chappaqua，实际运行的SQL如下：

       SELECT 
        places.CatID,
        places.PlaceName,
        places.PlaceID,
        places.Address1,
        places.PostalCode,
        places.Locality,
        places.Address2,
        places.ImageThumb,
        places.Adm1,
        places.Country,
        places.Adm2,
        places.Adm3,
        places.contributorid,
        places.lng,
        places.lat,
        places.verified,
        places.verified_by,
        places.verified_date
    FROM places INNER JOIN places_cats ON places.PlaceID = places_cats.PlaceID
    WHERE  

    (((PlaceName LIKE '%chappaqua%') OR (Country LIKE '%chappaqua%') OR (Adm1 LIKE '%chappaqua%') OR (adm2 LIKE '%chappaqua%') OR (adm3 LIKE '%chappaqua%') OR (locality  LIKE '%chappaqua%')))

    AND places_cats.CATID IN (1,21,15,32,16,26,29,27,28,25,75,89,38,5,36,88,87,31,33,24,35,37,90,39,40,34,30,9,8,7,11,20,19,96,97,95,13,17,14,12,3,2,4,84,85,86)
    GROUP BY 
        places.CatID,
        places.PlaceName,
        places.PlaceID,
        places.Address1,
        places.PostalCode,
        places.Locality,
        places.Address2,
        places.ImageThumb,
        places.Adm1,
        places.Country,
        places.Adm2,
        places.Adm3,
        places.contributorid,
        places.lng,
        places.lat,
        places.verified,
        places.verified_by,
        places.verified_date
        ORDER BY PlaceName

我知道它很丑陋而复杂。 运行大约需要1836毫秒。 有没有更好的方法来编写查询或代码，以便在一秒钟内加速返回的数据？

这是关于SQL的说明：

Answer 1

问题是引擎正在对表进行全表扫描，然后对group by排序。

全表扫描几乎是必要的，因为like的：

(((PlaceName LIKE '%chappaqua%') OR (Country LIKE '%chappaqua%') OR (Adm1 LIKE '%chappaqua%') OR (adm2 LIKE '%chappaqua%') OR (adm3 LIKE '%chappaqua%') OR (locality  LIKE '%chappaqua%')))

问题是无法使用PlaceName的索引，因为初始字符不是固定的。

所以。 。 。 你可以删除该group by吗？ 你至少可以用distinct替换它，虽然我认为这不会影响查询计划。 你有关于place_cats(placeId, catId)的索引吗？ 这至少会阻止查询读取类别表，因此它可以只进行索引查找。

你能将搜索限制在字段开头的单词吗？

我能想到的唯一另一种选择是切换到在MySQL中使用全文索引。