插1000条数据，需要22s，代码如下, columns是随机生成的一组数，

            df = pd.DataFrame()
            t1 = 0
            t2 = 0
            for i in range(1, num):
                st = time.time()
                columns = db_generator_helper.gen_table_columns(table_json)
                t1 += time.time() - st
                st = time.time()
                df = df.append(pd.Series(columns), ignore_index=True)
                t2 += time.time() - st
                if i % 1000 == 0:
                    df.to_csv(writer, encoding="utf-8")
                    print("write %d records to %s , gen data cost %f s, gen
DataFrame cast %f s" % (
                        i, table_name_without_db, t1, t2))
                    df = pd.DataFrame()
                    t1 = t2 = 0
--
修改:qlogic FROM 119.139.199.*
FROM 119.139.199.*

直接操作list算了
性能提升一倍，本来想偷懒用to_csv的功能

【 在 hgoldfish (老鱼) 的大作中提到: 】
: 你先放在 dict 里面，然后一次性 load 进去啊。
--
FROM 119.139.199.*