[mysql]python3实现并发访问水平切分表

场景说明

假设有一个mysql表被水平切分，分散到多个host中，每个host拥有n个切分表。 如果需要并发去访问这些表，快速得到查询结果， 应该怎么做呢？ 这里提供一种方案，利用python3的asyncio异步io库及aiomysql异步库去实现这个需求。

代码演示

import logging import random import asyncio from aiomysql import create_pool # 假设mysql表分散在8个host, 每个host有16张子表 TBLES = { "192.168.1.01": "table_000-015", # 000-015表示该ip下的表明从table_000一直连续到table_015 "192.168.1.02": "table_016-031", "192.168.1.03": "table_032-047", "192.168.1.04": "table_048-063", "192.168.1.05": "table_064-079", "192.168.1.06": "table_080-095", "192.168.1.07": "table_096-0111", "192.168.1.08": "table_112-0127", } USER = "xxx" PASSWD = "xxxx" # wrapper函数，用于捕捉异常 def query_wrapper(func): async def wrapper(*args, **kwargs): try: await func(*args, **kwargs) except Exception as e: print(e) return wrapper # 实际的sql访问处理函数，通过aiomysql实现异步非阻塞请求 @query_wrapper async def query_do_something(ip, db, table): async with create_pool(host=ip, db=db, user=USER, password=PASSWD) as pool: async with pool.get() as conn: async with conn.cursor() as cur: sql = ("select xxx from {} where xxxx") await cur.execute(sql.format(table)) res = await cur.fetchall() # then do something... # 生成sql访问队列, 队列的每个元素包含要对某个表进行访问的函数及参数 def gen_tasks(): tasks = [] for ip, tbls in TBLES.items(): cols = re.split('_|-', tbls) tblpre = "_".join(cols[:-2]) min_num = int(cols[-2]) max_num = int(cols[-1]) for num in range(min_num, max_num+1): tasks.append( (query_do_something, ip, 'your_dbname', '{}_{}'.format(tblpre, num)) ) random.shuffle(tasks) return tasks # 按批量运行sql访问请求队列 def run_tasks(tasks, batch_len): try: for idx in range(0, len(tasks), batch_len): batch_tasks = tasks[idx:idx+batch_len] logging.info("current batch, start_idx:%s len:%s" % (idx, len(batch_tasks))) for i in range(0, len(batch_tasks)): l = batch_tasks[i] batch_tasks[i] = asyncio.ensure_future( l[0](*l[1:]) ) loop.run_until_complete(asyncio.gather(*batch_tasks)) except Exception as e: logging.warn(e) # main方法, 通过asyncio实现函数异步调用 def main(): loop = asyncio.get_event_loop() tasks = gen_tasks() batch_len = len(TBLES.keys()) * 5 # all up to you run_tasks(tasks, batch_len) loop.close()