批量更新数据库技巧-优快云博客

下面这个和我的很相似，不用自己写了，就用他的吧！

最近做了一个需求，需要批量更新数据库表，但是因为涉及到的数据较多(千万条)，如果直接用sql更新，估计会把pl/sql弄垮

sql如下：update online_product set online_flag = '0' where status = 'ON'

所以，写了一个存储过程，以备忘：

declare
  cursor product_id_list is
    select product_id from online_product where status = 'ON';
  commit_count number := 0;
  total_count number := 0;
begin
  for pid in product_id_list loop
    total_count := total_count + 1;
    commit_count := commit_count + 1;
    update online_product setonline_flag = '0'
     where status = 'ON'
     and product_id = pid.product_id;
    if commit_count > 1000 then
      commit;
      commit_count := 0;
    end if;
  end loop;
  commit;
  dbms_output.put_line('Total Count:' || total_count);
end;

以 mod 判断提交更为简便（每500条提交一次）。

declare
    v_count number := 0;
    cursor cur_r_meas_basic_info is 
           select  trade_flow_no 
           from R_MEAS_BASIC_INFO i 
           where stake_no='1110102010002418' and trad_time<(select to_date('2016-05-19','yyyy-MM-dd') from dual);  
  begin
    for i in cur_r_meas_basic_info loop
       update R_TRADE_REC t set t.batch_no ='111' where t.trade_flow_no =i.trade_flow_no and t.batch_no is null;
            v_count:= v_count+1;
            if mod (v_count,500) = 0 then  
            commit;
            end if;
    end loop;
  commit;
  end;