Memcached vs. Redis

When working with dynamic and demanding ap­pli­ca­tions that use database con­nec­tions, opting for an in-memory database like Memcached or Redis sig­nif­i­cant­ly boosts per­for­mance. Both solutions are open-source and work in a similar way. However, when it comes to features and use cases, there are some dif­fer­ences that you should be aware of.

Put simply, Memcached is designed to be easy to use, whereas Redis offers a more com­pre­hen­sive range of features (whether they are helpful will depend on the use case). Let’s take a closer look at these two in-memory data stores and their dif­fer­ences and sim­i­lar­i­ties.

Firstly, Memcached has been around for almost 20 years, whereas Redis arrived on the market about 7 years later. The de­vel­op­ers of Memcached wanted to find a way of speeding up access to the kind of databases that are required for advanced web ap­pli­ca­tions – and they did just that. Their in-memory cache service (Memcached for short) uses key/value pairs to make cached data available at lightning speed, cir­cum­vent­ing the need for hard disk access.

Redis is short for Remote Dic­tio­nary Server. It too is a key-value in-memory database and, like Memcached, it is used for a wide range of projects, including websites, gaming, and online mar­ket­places.

The Memcached and Redis in-memory data stores read data at in­cred­i­bly fast speeds, reducing response times to just a few mil­lisec­onds or less. In terms of syntax, both systems are easy to use and can be in­te­grat­ed into ap­pli­ca­tions without any need for complex code. Moreover, both Memcached and Redis allow data to be dis­trib­uted across several nodes, so it is easy to scale systems ver­ti­cal­ly as the amount of data grows. Finally, they are open-source solutions and support numerous protocols and pro­gram­ming languages such as Java, Python, PHP, C, C++, C#, JavaScript, Node.js, Ruby, and Go.

Although Memcached and Redis have a lot in common, each has some unique features too. We will focus on the most important ones here.

Ar­chi­tec­ture: Memcached is multi-threaded, which means it can use multiple pro­cess­ing cores si­mul­ta­ne­ous­ly. Redis on the other hand uses a single core. This makes it more efficient at handling small amounts of data. For very large amounts of data, however, Memcached is the better option.

Data types: Memcached can only handle simple strings of up to 1 MB, whereas Redis supports values up to 512 MB in size and more complex data struc­tures like lists, hashes, streams, sets, sorted sets, bitmaps, and hy­per­loglogs. What’s more, Redis can also measure and store geospa­tial data.

Data per­sis­tence: Unlike Memcached, Redis is not simply an in-memory system. With Memcached, if there is no more space on a server for new items, the least recently used data will au­to­mat­i­cal­ly be deleted. Data is also lost if a Memcached instance fails. With Redis on the other hand, data can be restored using the “point-in-time” snapshot function or Append Only File (AOF) mode. Note however, that AOF mode may slow things down slightly.

Cluster man­age­ment: Redis uses both client-side and server-side caching to dis­trib­ute data, whereas Memcached only uses client-side caching. The nodes in a Redis cluster can exchange data with one another. There is no notable impact on response times if a node subset fails. In Memcached this kind of data exchange is not possible at all.

Trans­ac­tions: Both in-memory systems use what are known as “atomic commands”. This means that when values are entered, they are in­stan­ta­neous­ly displayed on the client side. In Redis, commands can also be grouped together in the form of isolated atomic op­er­a­tions. This feature, called pipelin­ing, means that Redis can respond to multiple commands at the same time. This is not the case in Memcached.

Pub/Sub: Redis supports the Pub/Sub ar­chi­tec­ture. This messaging pattern fa­cil­i­tates efficient com­mu­ni­ca­tion between ap­pli­ca­tions. It is used for high-per­for­mance chat rooms, live streams, social media feeds, and com­mu­ni­ca­tion between different servers.

Lua scripting: If de­vel­op­ers want to use Redis to perform complex cal­cu­la­tions, they can use Lua scripts. Lua is a light­weight pro­gram­ming language into which new logic can easily be added. This boosts per­for­mance and sim­pli­fies ap­pli­ca­tions.

So, which is best, Memcached or Redis? Both solutions are mature in-memory data stores. They differ mainly in terms of supported data formats, cluster man­age­ment, memory man­age­ment and data per­sis­tence. As for which engine is better, it depends on how you intend to use it. If you are already using Memcached and don’t need any extra func­tion­al­i­ty, then it may not be worth switching to Redis. However, if you do want more features and better backup options, then Redis is a good al­ter­na­tive.