When simple models supported by Memcached just won't do, we can use Redis. Redis is an in-memory data repository that can be used not only as a caching environment but also as a persistent database. It also has the ability to deliver message brokering with a publisher/subscriber model, allowing us to deliver services that do much more than just simple caching. While Memcached supports just simple values to be stored in the data structure, Redis is capable of storing advanced datasets such as strings, hashes, lists, sets, bitmaps, geospatial indexes, and so on, and supports performing complex transactions on the data with the Lua scripting language. It also supports the ability to create a cluster with one read/write node and up to five replicas. This means that, when there are a lot of concurrent writes from multiple nodes or multi-threaded applications, Redis might not be the correct choice, but it can still perform spectacularly well under heavy read load from multiple targets at once. Redis can also be deployed as a Multi-AZ cluster, making sure the data is replicated synchronously across availability zones for applications that require highly available caching, in-memory databases, or very fast messaging solutions.