If you need to store and process large amounts of data, cloud storage systems (SANs) are a good choice. There are no restrictions on the type of information; you can work with the usual data formats – documents, media files, archives, etc., as well as structured and unstructured data. The choice of storage type depends on the tasks, but storage systems generally do not differ much in their basis: they use SSD drives (All Flash Array) or hybrid solutions, i.e. SSD + HDD.
Block storages
Physically it can be several storages of different type, but the operating system perceives it as a logical drive, in relation to which all the traditional actions are available – it can be formatted, partitioned and the operating system installed. In accordance with its name, data is not stored here in the form of files, but in the form of blocks. Thanks to this, the data throughput (IOPS) is higher than other types of storage systems.
Such systems are suitable for the following:
Large “heavy” databases. SAP Hana, Oracle, etc.
Virtualization. Virtual machines usually use the disk space of block data storage.
High performance computing.
As storage for development environments.
Typically, block storage systems are used in SAN (Storage Attached Network) networks.
The disadvantages are the high cost. Also, the amount of metadata is limited.
File Storage
This type of storage is directories with files. The main advantage is its simplicity and it is cheaper than a block storage system. This topology is especially good if you don’t need to handle large amount of data.
Typically, such storage systems are built as follows:
File sharing systems for a distributed team to work on a project.
Local archiving systems for regular backups.
The main disadvantage is that when there is a lot of data, the speed of storage becomes slower. Therefore, they are best suited for situations where responsiveness is not critical. NAS (Network Attached Storage) file servers are usually built on such storage.
Object Storage.
Large amounts of data can be stored here – up to petabytes. They are not stored as files but as objects. They have unique identifiers and metadata, so it is easy to access objects. They also have internal mechanisms to check the integrity and availability of files. With detailed object metadata, you can optimize storage and storage costs.
Such repositories are used:
for machine learning and analytics: you can use them to implement new solutions and optimize cloud applications;
large media storage – for example, if you need to store a lot of high-resolution photos of products;
big data – for example, if your customers generate a lot of data and you want to analyze it;
backup systems;
cloud development;
hosting.
It is on object storage that the infrastructure of services such as Facebook, Dropbox, Netflix, Spotify is built. All of them appreciated the scalability, extensibility of metadata and speed of access to information. But that doesn’t make such storage a one-size-fits-all solution. So for your project, you have to choose between all options.
