我们最近已在Magento商店中进行了一些更改,从而触发了以下SQL的运行:
SELECT `count_table`.`category_id`, COUNT(DISTINCT count_table.product_id) AS `product_count`
FROM `catalog_product_flat_1` AS `e`
INNER JOIN `catalog_category_product_index` AS `cat_index` ON cat_index.product_id=e.entity_id AND cat_index.store_id=1 AND cat_index.visibility IN(2, 4) AND 1
INNER JOIN `catalog_product_index_price` AS `price_index` ON price_index.entity_id = e.entity_id AND price_index.website_id = '1' AND price_index.customer_group_id = 0
INNER JOIN `catalog_product_index_eav` AS `brand_idx` ON brand_idx.entity_id = e.entity_id AND brand_idx.attribute_id = '135' AND brand_idx.store_id = 1 AND brand_idx.value IN('967')
INNER JOIN `catalog_product_flat_1` ON catalog_product_flat_1.entity_id=e.entity_id
INNER JOIN `catalog_product_flat_1` AS `catalog_product_flat_1_2` ON catalog_product_flat_1.entity_id=e.entity_id
INNER JOIN `catalog_category_product_index` AS `count_table` ON count_table.product_id = e.entity_id
WHERE (e.status = 1) AND (catalog_product_flat_1.brand = '967') AND (catalog_product_flat_1.brand = '967') AND (count_table.store_id = 1) AND (count_table.category_id IN ('335', '334', '332', '339', '337', '943'))
GROUP BY `count_table`.`category_id`;
此SQL需要花费几秒钟的时间来运行,并且如果大量用户一次访问同一页面,则服务器最终将在查询备份时陷入停顿状态。
运行EXPLAIN
可以做到这一点:
1 SIMPLE brand_idx ref PRIMARY,IDX_CATALOG_PRODUCT_INDEX_EAV_ENTITY_ID,IDX_CATALOG_PRODUCT_INDEX_EAV_ATTRIBUTE_ID,IDX_CATALOG_PRODUCT_INDEX_EAV_STORE_ID,IDX_CATALOG_PRODUCT_INDEX_EAV_VALUE IDX_CATALOG_PRODUCT_INDEX_EAV_VALUE 4 const 17 Using where; Using index; Using temporary; Using filesort
1 SIMPLE count_table ref PRIMARY,IDX_CAT_CTGR_PRD_IDX_PRD_ID_STORE_ID_CTGR_ID_VISIBILITY,15D3C269665C74C2219037D534F4B0DC IDX_CAT_CTGR_PRD_IDX_PRD_ID_STORE_ID_CTGR_ID_VISIBILITY 6 db.brand_idx.entity_id,const 1 Using where; Using index
1 SIMPLE e eq_ref PRIMARY,IDX_CATALOG_PRODUCT_FLAT_1_STATUS PRIMARY 4 db.brand_idx.entity_id 1 Using where
1 SIMPLE cat_index ref IDX_CAT_CTGR_PRD_IDX_PRD_ID_STORE_ID_CTGR_ID_VISIBILITY,15D3C269665C74C2219037D534F4B0DC IDX_CAT_CTGR_PRD_IDX_PRD_ID_STORE_ID_CTGR_ID_VISIBILITY 6 db.brand_idx.entity_id,const 1 Using where; Using index
1 SIMPLE catalog_product_flat_1 eq_ref PRIMARY PRIMARY 4 db.brand_idx.entity_id 1 Using where
1 SIMPLE price_index eq_ref PRIMARY,IDX_CATALOG_PRODUCT_INDEX_PRICE_CUSTOMER_GROUP_ID,IDX_CATALOG_PRODUCT_INDEX_PRICE_WEBSITE_ID,IDX_CAT_PRD_IDX_PRICE_WS_ID_CSTR_GROUP_ID_MIN_PRICE PRIMARY 8 db.brand_idx.entity_id,const,const 1 Using index
1 SIMPLE catalog_product_flat_1_2 index NULL IDX_CATALOG_PRODUCT_FLAT_1_ATTRIBUTE_SET_ID 2 NULL 21529 Using index; Using join buffer (flat, BNL join)
对我来说,这暗示使用brand_idx
的{{1}}表基于某处缺少索引。但这是真的吗?
如果是的话,如何确定丢失的索引,其次,我将如何在Magento中应用该索引?
我知道这还与选择Using temporary; Using filesort
并按COUNT(DISTINCT count_table.product_id)
分组。count_table
有关,因为删除这两个部分中的任何一个都会导致查询更快(只是没有预期的信息!)。
答案 0 :(得分:0)
此问题是由于两个inner join
引用同一表catalog_product_flat_1
的结果,但是第二个连接实例的连接条件从第一个实例引用了该表:>
INNER JOIN `catalog_product_flat_1` ON catalog_product_flat_1.entity_id=e.entity_id
INNER JOIN `catalog_product_flat_1` AS `catalog_product_flat_1_2` ON **catalog_product_flat_1**.entity_id=e.entity_id
尽管该问题涉及Magento,但这实际上与Zend_Db和定义联接的方式有关。
鉴于Zend_Db_Select对象,您可以像这样创建联接:
$select->joinInner('catalog_product_flat_1', 'catalog_product_flat_1.entity_id = e.entity_id');
第一次执行此操作是可以的,但是如果第二次执行此操作,则joinInner
函数将智能地标识catalog_product_flat_1
表上的第二个联接并将其命名为{{ 1}},但是会出现此问题,因为它无法在联接条件中识别出错误的表别名。
解决此问题的方法是通过提供名称相关性来显式设置表别名,如下所示:
catalog_product_flat_1_2
在第二个联接引用正确的表别名的情况下,查询时间从$select->joinInner(array('unique_table_alias' => 'catalog_product_flat_1'), 'unique_table_alias.entity_id = e.entity_id');
减少到3s
。
Magento两次添加内部联接的事实是另一回事,但是一旦配置正确,就不会影响性能。