这2个查询语句怎么合为一个？

[firmbank]

有三张表 aa, bb, cc
表字段均一样 (id, flg)

select a.id, b.flg from aa a, bb b where a.flg = '1' and a.id = b.id
union
select a.id, c.flg from aa a, cc c where a.flg != '1' and a.id = c.id;

aa表数据量很大且flg非索引字段，bb和cc很小，怎么优化这个语句？
谢谢！

[sunboy_yf]

指定bb，cc为驱动表试验一下。

[_拙计]

select flg, id
          from bb
         where exists (select 1
                  from aa
                 where aa.flg = '1'
                   and aa.id = bb.id)
        union all
        select flg, id
          from cc
         where exists (select 1
                  from aa
                 where aa.flg <> '1'
                   and aa.id = cc.id)

会不会快点？