What I like best about Redis is that it is fast. Absolutely, blazingly fast. I primarily use it as an edge cache on my applications for data that I would not want to do a round trip to either Mongo or MySQL for and I am constantly blown away by how quick it responds. It's like memcache on steroids.
The biggest thing I dislike about redis is that without proper management one can end up with memory fragmentation issues that cause performance degradation when writing large amounts of data. Granted, I haven't run into this issue as most of what I have used Redis for is profile and product data - records ranging in size from 1kb to 10kb at the largest.
Try it, at least once - I'm sure you'll fall in love with it. If you're worried about persistency related performance degradation, a few tweaks to the kernel are all that need be done. Nothing painful!
On the small end, I have used Redis to considerably increase the performance of client applications that were largely bloated and did many back and forths to the database as they were architected. When the client brought me on to fix their performance issues, the first thing I did was plug Redis in as a front data cache. Additionally, I have used Redis for a personal application with ephemeral data, using it as a simple KV store for the interim results of certain steps of a long running calculation.