Database Reference
In-Depth Information
DynamoDB versus Redis
Both DynamoDB and Redis are NoSQL databases that store the data in key-value format.
But Redis is an open source database provided by BSD.
Redis
means
Remote Dictionary
Server
. It is often called a data structure server as its key contains many data types, such as
string hashes, in which the key and values are string, sorted sets of strings, stored sets, and
lists of strings. We can perform atomic operations in Redis. Redis stores the whole dataset
in memory, so we can persist it by dumping the data to disk, because Redis synchronizes
data to the disk every 2 seconds. So if the system fails, we lose the data for only a few
seconds. Another way to persist it is by appending each command to the log.
Redis supports master-slave replication, which allows the slave Redis server to be the same
copy of the master Redis server with non-blocking replication. On the master-side server, it
will handle queries when one or more slaves perform initial synchronization. On the other
side, while the slave server performs initial synchronization, it will still handle queries us-
ing the old dataset. But in DynamoDB, replication is done at a scheduled time; a continu-
ous data replication is not performed. So in this case, if a primary DynamoDB table loses
its data, there can be data loss while restoring from the backups. In Redis, slaves, are able
to accept connections from several slaves apart from slaves from the same master server.
But in DynamoDB, the tables should be on the same AWS account.
Other features of Redis include transactions, pub/sub, Lua scripting capabilities, keys with
a restricted time to live, and other configuration settings, which will allow Redis to work
like a cache. Redis is written in ANSI C and works in almost all operating systems, such as
Linux, BSD, and OS X without any external dependencies.
While data durability is not the major concern, the in-memory environment of Redis allows
it to perform extremely well compared to database systems that write every update or
change to disk before allowing a committed transaction. There is no prominent speed dif-
ference between read and write IOs. Redis works as a single process and is single-threaded.
Hence a single Redis instance cannot apply parallel execution of tasks such as stored pro-
cedures. Redis is mainly used for rapidly changing data with a predictable database size
that should mostly fit in-memory. So Redis is used in real-time applications such as storing
real-time stock prices, real-time communication, real-time analytics, and leaderboards. We
can use it as an option for memory cache too. Let's move on to the tabular comparison
between these databases, as follows:
