get performance

Hi, I am processing large dataset and it looks like I am struggling with redis look up performance. there are around 600M keys. 4 MGET calls take more than 40ms in multithreaded (50 threads) env, and redis connection pool (50 connections) on my mac book pro. I am using Jedis, btw.

Every MGet is fetching results for 100 keys of the same pattern (as I read somewhere on forum that is recommended number).

I'd like to know if I can pipeline those 4 MGET calls, so i get some performance gain?
I couldn't find how to do that using RedisTemplate and Jedis. Any hint?

Also are there any significant differences in speed among available drivers (jedis, jredis etc). Is there any benchmark available?

Thanks in advance,
Milan