connection pooling

[tanner0101]

Allow connections to be pooled and re-used.

[John-Connolly]

I think pooling would only be useful with pub/sub or blocking commands, which is probably not the most common scenario. It would be nice if this was opt in, so it doesn't exhaust connection limits on cloud providers.

[jdmcd]

Exhausting connection limits on cloud providers is actually what I was worried about. I use Redis to store session tokens and I don't my whole massive setup to come down just cause the caching service is out of connections (and is only used on session lookup). I see what you're saying though. I'd love to see a middleground

[tanner0101]

Maybe we can have it configurable such that if you set max connections to ‘nil’ it will just create a new connection each time. Note also the connection pooling is for the KeyedCache impl. Creating a Redis client manually will not be affected by this.

[Joannis]

I don't see a reason to pool redis connections since they support pipelining. There's little to no performance gain. Actually, I think it'll worsen the performance very slightly.

[John-Connolly]

Yeah the only way I see it being useful is for blocking commands. I believe blocking commands put the client in a blocked state and the client is unable to send or receive other commands until that blocking call returns or times out.

[Joannis]

I agree, but we don't support blocking operations in the Redis Client. We have a separate static method for setting up a blocking client for pub/sub.

[John-Connolly]

I see, creating a new RedisClient instance for those commands makes more sense anyways. The only problem I see is there is no way to send commands before subscribe is sent, like AUTH or SELECT for instance.

[tanner0101]

Fixed by #95