如何改善python-kafka的消费

Question

我测试了我的python-kafka脚本，kafka-producer产生了10000 / s的消息，但我耗费了大约1000 / s，那么如何在整个过程中改善我的消费者呢？

while True:
    topic_to_records = consumer.poll(100)
    if len(topic_to_records.values()) == 0:
        time.sleep(0.05)
    for records in topic_to_records.values():
        print 'consume %s messages' % len(records)
        for msg in records:
            print (time.strftime('%Y-%m-%d %H:%M:%S', time.localtime()), '*'+msg.value.split('\t')[0])