I'm wondering why redis-rb doesnt support connection pooling internally?

Redis::Client already appears to maintain the handle to the actual driver/connection.

We use redis-rb heavily and are looking into fibers + async IO and have connections to multiple redis servers.

As I understand how this works is that redis/connection/synchrony will give me non-blocking IO and fiber aware driver.

if I'm running multiple fibers and only 1 instance of redis-rb per redis server, I don't see how using this driving will give me any kind of real concurrency without a fiber aware connection pool.

So a connection pool seems to be the obvious solution.

I also want/need it to be transparent to the code. em-synchrony/connection_pool seems like the obvious choice in my case, but the execute API makes integration a major pain and lots of code changes.

Thanks.
-Peter