Why NoSQL?

Why NoSQL?

Relational databases have become part of the computer culture due to their many advantages:

Obtain persistent data

Data can be persisted to the disk, and it is easy to access a small piece of data.

Concurrency

Relational databases have the concept of "Transactions", which effectively ensures consistency when multiple programs access the same data. ACID (atomicity, consistency, isolation, and durability) of transactions) I believe everyone is familiar with the features.

Integration

Applications of different development teams, platforms, and languages can share the same data.

Standard relational model

The reason for achieving this goal is that a standard and unified relational model is provided.

However, two reasons make relational databases inadequate:

Impedance detuning

The data structure in the memory is very different from the relational model, which is called impedance mismatch )". Memory structures such as "nested records" and "list" must be converted to "Links" and "tuples" before they can be stored.

Kubernetes Clusters

With the sharp increase in the network scale, the "horizontal expansion" of hardware computing resources has become inevitable, and clusters have become a trend. However, relational databases are not designed for clusters, either the "shared Disk Subsystem" is used to support the file system on the cluster (the disk will become a bottleneck), or the database is split according to the logic, although the load can be distributed to multiple servers, however, the application must control all shards, and transactions and consistency cannot be performed across shards.

So far, NoSQL was born.

• Created for clusters

Basically, NoSQL is born for clusters (not the case for a few databases). When the amount of data to be processed is large or increases rapidly, it also requires high data access efficiency, consider NoSQL.

• Impedance detuning

Remember that processing "Big Data" on clusters is not the only reason for NoSQL, and impedance detuning is another key factor. NoSQL databases do not have the concept of "pattern" in traditional relational databases, without the need to define a data storage model, you can add fields at will, which is very suitable for processing irregular data and custom fields, when you want to improve application development efficiency with a more convenient and friendly data interaction method, consider NoSQL.