优化此SQL查询

Question

这个SQL查询让我感到厌恶.我没有写它,但它是我们服务器问题的一个重要原因.我愿意将它分成多个查询并通过PHP进行一些处理(比如,RAND()).

$sql = "SELECT a.code, a.ad_id, a.position, a.type, a.image, a.url, a.height, a.width
    FROM " . AD_TABLE ." a, " . USER_GROUP_TABLE . " g
    WHERE (a.max_views >= a.views OR a.max_views = '0')
    AND (FIND_IN_SET(" .$forum_id. ", a.show_forums) > 0 OR a.show_all_forums = '1')
    AND g.user_id = " . $user->data['user_id'] . "
    AND FIND_IN_SET(g.group_id, a.groups)
    AND FIND_IN_SET(" . $user->data['user_rank'] . ", a.ranks)
    AND a.start_time < " . time() . "
    AND a.end_time > " . time() . "
    AND (a.clicks <= a.max_clicks OR a.max_clicks = '0')
    ORDER BY rand()";

是的,粘贴之后我觉得很蠢...

编辑:以下是上述格式的示例查询"EXPLAIN"的结果,逗号分隔:

"id","select_type","table","type","possible_keys","key","key_len","ref","rows","Extra"
1,"SIMPLE","g","ref","user_id","user_id","3","const",6,"Using temporary; Using filesort"
1,"SIMPLE","a","ALL","max_views","","","",10,"Using where"

它是

Answer 1

你有三个主要问题:

1. FIND_IN_SET.

   这不是一件容易的事,一个索引不能让它变得更快.创建多对多关系表(或多个表).

   • a.start_time < GETDATE() AND a.end_time > GETDATE()

   MySQL优化它并不好.您可以将时间跨度保存为几何框并SPATIAL INDEX在其上创建,这将更快(尽管可读性较差)

   • ORDER BY RAND()

   如果您使用它来对数据进行采样(即您不需要所有行而是需要一个小的随机子集),那么有一种更有效的方法可以执行此操作,如本文在我的博客中所述:

   • 选择随机行