Comment 10 for bug 1578401

Re-reading the existing implementation I think this is a valid improvement for memcache driver and also redis driver, which do not currently use the expiration mechanism within backend service.

I'll look into this during this cycle, but one thing I noticed here is the fact that the solution earlier proposed breaks the case where a specific cache_time option is configured by a very large value. Probably we have to add a new option to control expiration in backends.