What is Redis?

Relational databases are used in most IT projects. Regardless of whether they are used for web design or software development, table-based databases tend to deliver solid, stable results in most situations. But are they always the perfect choice? In recent years, an increasing number of new database types have come onto the market, each with their own unique advantages. NoSQL databases are highly recommended for certain projects. Redis supplies such a database. How does the data storage work and what are the advantages of Redis?

The name Redis stands for Remote Dictionary Server. This type of server is designed for high-speed data storage. As a database management system (DBMS), Redis offers both an in-memory database and a key-value store.

• In-memory database: In these databases, the DBMS stores all data directly in the memory. This results in very fast access times, even for large amounts of unstructured data.
• Key-value store: Key-value databases are also impressive with their high performance and easy scalability due to their simple structure. A key is created for each entry which can then be used to retrieve the information.

So, all the data in a Redis server is not stored on the hard drive but in the memory instead. As a result, Redis is both a cache and a memory: it makes no difference for Redis whether the information is stored permanently in the database or only for a short time.

Every entry in the database is assigned a key. Using these keys, data can be easily retrieved. Entries are not linked to one another and thus do not need to be requested across multiple tables. The information is directly available.

Storing data in the memory also means that there is a risk of losing all the data if the server crashes. To safeguard against this eventuality, Redis can either copy all the data regularly to a backup hard drive or save all the commands needed for reconstruction in a log file.

Strings are the common data structure for Redis (i.e. simple character strings). Even keys (part of the key-value store) are strings. However, this system can also handle other data structures:

• Strings: a string of characters with a maximum size of 512 MB
• Hashes: an entry with multiple fields
• Lists: a collection of strings sorted based on the order they were entered
• Sets: an unsorted collection of strings
• Sorted sets: a user-sorted collection of strings
• Bitmaps: a collection of bit-level operations
• HyperLogLogs: an estimation based on unique values
• Streams: a list of strings or key-value pairs

Since Redis is open source, many developers are working on extensions for this DBMS. These modules add more functionality to the otherwise very simple database and adapt the software for specific applications.

The standard application for Redis is a cache. For example, Twitter uses the database for this purpose. The timeline in this messaging service is implemented using a Redis cache. Basically, the timeline consists of a list which can be quickly retrieved and incrementally extended thanks to Redis.

Redis is also used for other purposes. For example, Redis databases are perfect for messaging services and chat rooms due to their high access speeds. New entries can be displayed in real time. Even lists which must be quickly modified are implemented using Redis.

Redis is always a good choice if you need simple data to be available quickly. Due to the fact that it makes standard data backups, the read speed is increased: data can be retrieved from different instances. Scaling also works well when using Redis, both horizontally and vertically. If the Redis server’s memory is insufficient, the software has built-in virtual memory management. This stores data on the hard drive. Redis itself is written in C, but there are clients for almost every programming language.

However, if you are working with complex data which requires equally extensive query options, Redis is not the right choice. Generally, data in a key-value store can only be accessed via its keys. In addition, Redis (like other in-memory databases) requires an extremely large amount of memory which can be expensive. You should also expect to provide more memory than the data actually uses.