我有一个记录设备数据连接的表，该表目前为8200万行。我有一个nother表，其中包含设备信息，如位置，序列号，客户等，它有4000条记录。
表1具有以下结构：
+---------------+-------------+
| Column | type |
+---------------+-------------+
| id | int(11) | Primary Key
| billing_date | date |
| imsi | varchar(255)|
| bytes_input | int(10) |
| bytes_output | int(10) |
| total_in_kibs | int(10) |
+-----+---------+-------------+
：group_idx，包括imsi，billing_date，total_in _kibs，bytes_input，bytes_output
表2具有以下结构：
+---------------+-------------+
| Column | type |
+---------------+-------------+
| id | int(11) | Primary Key
| System | varchar(50) |
| Customer | varchar(50) |
| Sitecode | varchar(50) |
| Serialnumber | varchar(50) |
| Name | varchar(50) |
| imsi | varchar(255)|
+-----+---------+-------------+
索引：site_idx，包括imsi，名称，系统，客户，站点代码，序列号 id是主键
我想要做的是找到每个设备在给定月份内使用的总数据量total_in_kibs列。
我使用的查询是：
select x.imsi,y.name, y.Customer, y.System, y.Serialnumber,
(x.bytesin / 1048576) as 'In (Mb)', (x.bytesout / 1048576) as 'Out (Mb)',
(x.total / 1024) as 'Total (Mb)'
FROM
(SELECT imsi, sum(bytes_input) as bytesin,
sum(bytes_output) as bytesout, sum(total_in_kibs) as total
FROM table1
WHERE month(billing_date) = 5
GROUP BY imsi) as x
JOIN table2 as y
on y.imsi = x.imsi
explain命令提供以下内容：
+---+-----------+------------+------+---------------+----------+---------+------------+-------------------------------------+
| id|select_type| table | type | possible_keys | key | key_len | ref | rows | Extra |
+---+-----------+------------+------+---------------+----------+---------+------------+----------+--------------------------+
| 1 | PRIMARY | y | index | imsi,sign_idx| sign_idx | 479 | NULL | 4100 | Using where; Using index |
| 1 | PRIMARY | <derived2> | ref | key0 | key0 | 258 | data.y.imsi| 20087 | |
| 2 | DERIVED | table1 | index | NULL | group_idx| 272 | NULL | 82358731 | Using where; Using index |
+---+-----------+------------+------+---------------+----------+---------+------------+----------+--------------------------+
有没有办法加快查询速度？因为它将被用作网页上的搜索工具的一部分，因此页面加载的时间有点多。
先谢谢了 - 詹姆斯
Table 1
+-----------------------+-------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-----------------------+-------------+------+-----+---------+----------------+
| id | int(11) | NO | PRI | NULL | auto_increment |
| billing_date | date | NO | MUL | NULL | |
| unique_id | char(10) | NO | | NULL | |
| imsi | bigint(15) | NO | | NULL | |
| tim_state | char(2) | NO | | NULL | |
| customer_profile_name | varchar(25) | NO | | NULL | |
| serving_opco | char(5) | NO | | NULL | |
| session_start | datetime | NO | | NULL | |
| session_end | datetime | NO | | NULL | |
| bytes_input | int(10) | NO | | NULL | |
| bytes_output | int(10) | NO | | NULL | |
| total_in_kibs | int(10) | NO | | NULL | |
+-----------------------+-------------+------+-----+---------+----------------+
Indxes for table 1
--+------------+------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+--------------+------------+------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| transactions | 0 | PRIMARY | 1 | id | A | 91714401 | NULL | NULL | | BTREE | | |
| transactions | 1 | billing_date_idx | 1 | billing_date | A | 18 | NULL | NULL | | BTREE | | |
+--------------+------------+------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
Table 2
+----------------------+-------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+----------------------+-------------+------+-----+---------+----------------+
| id | int(11) | NO | PRI | NULL | auto_increment |
| System | varchar(50) | YES | | NULL | |
| Customer | varchar(50) | NO | | NULL | |
| Sitecode | varchar(50) | NO | | NULL | |
| Name | varchar(50) | NO | | NULL | |
| Serialnumber | varchar(10) | NO | | NULL | |
| Operator | varchar(50) | NO | | NULL | |
| Sign_Serialnumber | varchar(10) | NO | | NULL | |
| imsi | bigint(15) | YES | MUL | NULL | |
| dtInstalledDateGMT | datetime | NO | | NULL | |
| dtUninstalledDateGMT | datetime | NO | | NULL | |
+----------------------+-------------+------+-----+---------+----------------+
Indexes for table 2
+-------+------------+----------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+-------+------------+----------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| signs | 0 | PRIMARY | 1 | id | A | 4172 | NULL | NULL | | BTREE | | |
| signs | 1 | imsi | 1 | imsi | A | 4172 | NULL | NULL | YES | BTREE | | |
| signs | 1 | sign_idx | 1 | imsi | A | 4172 | NULL | NULL | YES | BTREE | | |
| signs | 1 | sign_idx | 2 | Name | A | 4172 | NULL | NULL | | BTREE | | |
| signs | 1 | sign_idx | 3 | System | A | 4172 | NULL | NULL | YES | BTREE | | |
| signs | 1 | sign_idx | 4 | Customer | A | 4172 | NULL | NULL | | BTREE | | |
| signs | 1 | sign_idx | 5 | Sitecode | A | 4172 | NULL | NULL | | BTREE | | |
| signs | 1 | sign_idx | 6 | Serialnumber | A | 4172 | NULL | NULL | | BTREE | | |
+-------+------------+----------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
Explain for billing_date index query
+------+-------------+-------+------+------------------+----------+---------+---------------------+----------+----------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+------+-------------+-------+------+------------------+----------+---------+---------------------+----------+----------------------------------------------+
| 1 | SIMPLE | x | ALL | billing_date_idx | NULL | NULL | NULL | 91714401 | Using where; Using temporary; Using filesort |
| 1 | SIMPLE | y | ref | imsi,sign_idx | sign_idx | 9 | vodafonegdsp.x.imsi | 1 | Using index |
+------+-------------+-------+------+------------------+----------+---------+---------------------+----------+----------------------------------------------+
答案 0 :(得分：0)
如果表中的imsi是唯一的（怀疑它与表1中的结算日期相结合，并且在另一个表上可能是唯一的），那么您可以消除子查询： -
SELECT x.imsi,
y.name,
y.Customer,
y.System,
y.Serialnumber,
(SUM(x.bytes_input) / 1048576) AS 'In (Mb)',
(SUM(x.bytes_output) / 1048576) AS 'Out (Mb)',
(SUM(x.total_in_kibs) / 1024) AS 'Total (Mb)'
FROM table1 AS x
INNER JOIN table2 AS y
ON y.imsi = x.imsi
WHERE month(x.billing_date) = 5
GROUP BY x.imsi
另外，group_idx的表1上的索引（包括imsi，billing_date，total_in _kibs，bytes_input，bytes_output）似乎过多。我怀疑各种总数在该指数的末尾是有用的，只是通过imsi /结算日期缩小范围将非常有效。
修改
我正在努力寻找一种方法来强制它使用有用的索引。以下可能会做，但远非确定。
请尝试以下操作，但在billing_date（只是结算日期）上有一个索引： -
SELECT y.imsi,
y.name,
y.Customer,
y.System,
y.Serialnumber,
(SUM(x.bytes_input) / 1048576) AS 'In (Mb)',
(SUM(x.bytes_output) / 1048576) AS 'Out (Mb)',
(SUM(x.total_in_kibs) / 1024) AS 'Total (Mb)'
FROM table1 AS x
INNER JOIN table2 AS y
ON y.imsi = x.imsi
AND x.billing_date BETWEEN '2014/05/01' AND '2014/05/31'
GROUP BY y.imsi,
y.name,
y.Customer,
y.System,
y.Serialnumber
然后应该使用该索引返回table1中用于May的子行集。如果你在imsi上的table2上有索引（作为覆盖索引中的第一列，或索引中唯一的列），那么应该用于连接。
使用MONTH（billing_date）= 5将不使用索引，因为它必须在检查结果之前对行执行MONTH功能（8200万次）。检查日期范围应使用结算日期的索引。