使用同步和异步方式更新插入MongoDB数据的性能对比

import time
import pymongo
import motor.motor_asyncio
import asyncio
import random
# MongoDB 连接设置
MONGODB_HOST = 'localhost'
MONGODB_PORT = 27017
MONGODB_DB = 'test_db'
MONGODB_COLLECTION1 = 'test_collection1'
MONGODB_COLLECTION2 = 'test_collection2'
# 生成测试数据
def generate_test_data(num_records):
    return [{'asin': f'ASIN_{i}', 'data': random.random()} for i in range(num_records)]

# 同步插入数据的函数
def insert_data_sync(data):
    client = pymongo.MongoClient(MONGODB_HOST, MONGODB_PORT)
    db = client[MONGODB_DB]
    collection1 = db[MONGODB_COLLECTION1]
    start_time = time.time()
    for record in data:
        collection1.update_one({'asin': record['asin']}, {'$set': record}, upsert=True)
    end_time = time.time()
    client.close()
    return end_time - start_time

# 异步插入数据的函数
async def insert_data_async(data):
    client = motor.motor_asyncio.AsyncIOMotorClient(MONGODB_HOST, MONGODB_PORT)
    db = client[MONGODB_DB]
    collection2 = db[MONGODB_COLLECTION2]
    start_time = time.time()
    async def insert(record):
        await collection2.update_one({'asin': record['asin']}, {'$set': record}, upsert=True)
    await asyncio.gather(*[insert(record) for record in data])
    end_time = time.time()
    client.close()
    return end_time - start_time

# 主函数来运行测试
def main():
    data = generate_test_data(1000)  # 生成测试数据
    # 测试同步插入
    sync_time = insert_data_sync(data)
    print(f'Synchronous insertion time: {sync_time:.2f} seconds')
    # 测试异步插入
    loop = asyncio.get_event_loop()
    async_time = loop.run_until_complete(insert_data_async(data))
    print(f'Asynchronous insertion time: {async_time:.2f} seconds')
if __name__ == '__main__':
    main()

Synchronous insertion time: 27.93 seconds
Asynchronous insertion time: 6.84 seconds