Demystifying Cloud Storage
Data is the most important resource in any technological system, Data contains our business knowledge, customer base, and even our secrets. Therefore, choosing the correct storage solutions is critical to the success of any organization.
Unlike conventional data centers where all the storage resides on disks, the cloud has a rich variety of storage categories and each category has its price, speed, advantages, and disadvantages.
In this article, I will try to bring order in the cloud storage jungle and explain how to choose the best storage solution for every budget and business-technological need.
I divided the world of cloud storage into 3 storage families separated by the access method: *
1. REST API - Storage as a service
Storage is separated from compute
2.MOUNT - Drive
Storage is attached to compute as a filesystem
3.DRIVER - Database
Storage and compute are combined
* Within each storage family there are several storage categories. The list of categories is presented in front of you from the slowest and cheapest to the fastest and most expensive. The information is general and relevant to all types of clouds. Of course, there are exceptions in terms of features and prices.
Family 1: REST API based storage
- Relatively slow storage because storage is not "close" to servers and does not have the computing power for queries.
- Only possible to read complete files, that is, you can not read parts of files.
- It can be easily accessed from anywhere in the cloud and from outside.
- It can be accessed by multiple users.
- You can not mount this type of storage to a server
Categories in this family include:
- Archive storage
- Object storage
- CDN
Family 2: Mount-based
These storage types are accessible from servers as mounter internal drives.
Categories in this family include:
- Block storage
- Network file storage
Family 3: Driver-based storage - Databases
Databases are the most expensive form of storage.
This storage family is unique in that it includes a computational engine that enables the rapid retrieval of information by a query language. It is expensive because the calculation engine usually needs a lot of memory and CPU resources.
The categories in this family include:
- SQL Database
- NoSQL Database
- Data Warehouse
- In-memory Database
Summary
The choice of how to store data is an important and strategic architectural choice for the organization. Here are some rules of thumb for an efficient storage architecture:
- Keep backups and unused information in archive storage.
- To maintain low costs, the information should be stored in the cheapest storage level (object storage rather than on disks/databases).
- Information retrieved very frequently should be in SQL \ NOSQL \ REDIS databases.
- You should implement an “object lifecycle policy” that allows you to automatically migrate data from expensive storage categories to cheaper ones.
