redis workload player

redis-pipe-commands.py

import sys
import redis
import time

i = 0

conn = redis.Connection('localhost', 6379)

start_t = time.time()
for line in sys.stdin.xreadlines():
	args = line.split(" ")
	cmd = [c.replace('"', '') for c in args[3:]]
	conn.send_command(*cmd)
	i += 1
	if i % 1000 == 0:
		end_t = time.time()
		print "{} commands took {}s".format(i, end_t - start_t)
		start_t = end_t

Does conn.send_command block? Looks a bit like "coordinated omission" (http://www.azulsystems.com/sites/default/files/images/HowNotToMeasureLatency_LLSummit_NYC_12Nov2013.pdf) but it looks difficult to schedule something every X unit-of-time to avoid it.