Mysql Explain 查看 filtered 優化sql

sql

SELECT mt.* ,ds.ps_count FROM
        (select
         
        do.id,
        do.logistics_company_id,
        do.order_no,
        do.lable_info,
        do.ext_order_no,
        do.ext_order_no2,
        do.ext_po_no,
        do.owner_id,
        do.owner_name,
        do.shop_id,
        do.warehouse,
        do.warehouse_name,
        do.delivery_type,
        do.order_type,
        do.rowstate,
        do.receiver,
        do.province,
        do.city,
        do.country,
        do.address,
        do.mobile_no,
        do.merge_info,
        do.invoice_title,
        do.product_weight,
        do.expr_company,
        do.expr_bill_no,
        do.order_date,
        do.paid_date,
        do.picked_date,
        do.shipped_date,
        TIMESTAMPDIFF(HOUR ,nvl(do.paid_date,now()),nvl( do.shipped_date,now())) as deal_hours,
        do.buyer_note,
        do.seller_note,
        do.note_text,
        do.buyer_id,
        do.id_card_no,
        do.sender,
        do.sender_mobile,
        do.sender_province,
        do.sender_city,
        do.sender_country,
        do.sender_addr,
        do.pick_list_no,
        do.expr_printed,
        do.bill_printed,
        do.pick_type,
        do.part_count,
        do.part_list,
        do.ean_no_list,
        do.part_no_list,
        do.shortage_flag,
        do.pkg_weight,
        do.sync_flag,
        do.create_order_type,
        do.error_msg,
        do.created_date,
        do.created_user,
        do.updated_date,
        do.updated_user,
        do.cut_reason,
        do.goods_count,
        do.print_no,
        do.box_no,
        do.source_platform_code,
        do.ext_order_id
     
        from delivery_order do -- force index(delivery_order_6ix)
        where 1=1 AND do.logistics_company_id = 10000
        and do.owner_id  
         
            in
             (  
                'BGSS'
             , 
                'DBQ'
             , 
                'BSA'
             , 
                'ABJ'
             , 
                'CAT'
             , 
                'DBK'
             , 
                'TS'
             , 
                'JCX'
             , 
                'WCX'
             , 
                'FY'
             , 
                'WHN'
             , 
                'DDG'
             , 
                'YM'
             , 
                'PTJ'
             , 
                'MG'
             , 
                'WYM'
             , 
                'TJ'
             , 
                'TSMJ'
             , 
                'CXCW'
             , 
                'MYCW'
             , 
                'BK'
             , 
                'CXJD'
             , 
                'CXTM'
             , 
                'QCX'
             , 
                'HS'
             ) 
         
         
     
        and do.warehouse  
         
            in
             (  
                'WHC'
             , 
                'QHDC'
             , 
                'JKXS'
             , 
                'JSC'
             , 
                'WHCPC'
             , 
                'CXB'
             , 
                'WHN'
             ) 
         
            AND do.rowstate IN
             (  
                2
             ) 
                AND do.owner_id IN ('BGSS')
 
        )AS mt
        LEFT JOIN
        (SELECT do2.id,do2.part_list,ps.ps_count from delivery_order as do2
		        LEFT JOIN
		        (SELECT
				        part_list,
				        (CASE
				        WHEN (COUNT(*) > 0) THEN COUNT(*)
				        ELSE 0
				        END) AS ps_count FROM delivery_order WHERE rowstate = 2 GROUP BY part_list) AS ps
				        ON do2.part_list = ps.part_list  and  do2.rowstate = 2 
		   ) as ds
        ON mt.id = ds.id
        WHERE 1=1
        order by rowstate ASC ,created_date DESC;

 執行總用時 4秒多, 一個select count 一個 select ,形成一面刷新須要8秒多

使用explain 發現 filtered 爲0.1, 總數據量是 60多w

考慮到頁面加載的時候,經常使用查詢字段,增長聯合索引

logistics_company_id, owner_id, warehouse, rowstate

只須要 0.8秒多

filtered 爲 100 ,快了不是一個deng'ji