Distinctions Between Storage Modes and Storage Types

When it comes to the interface of their storage, there are three frequent application scenarios for distributed storage systems:

Object Storage

To elaborate, there are classic key-value storage options along with their associated commands such as GET, PUT, and DEL. Moreover, seven cow, another beat Swift and S3 have been added to the suite of available functions for your convenience.

Block Storage

Commonly, this interface is a QEMU Driver or Kernel Module. The QEMU-provided Linux Block Device or Block Driver interfaces, such as Sheepdog and AWS EBS, must be implemented by this interface. 

File Storage

Supporting POSIX interfaces, which are similar to traditional file systems like Ext4, is a common meaning. However, distributed storage provides parallelization capabilities, like CephFS of Ceph, though GFS and HDFS, which are non-POSIx file storage interfaces, can also be included in this category.

Block Storage 

The following are the two types of block storage:

Direct Attached Storage, or DAS: 

The host server is connected to the storage device in a direct manner, with each having its own independent memory. In order to access data from the mainframe, you must go through multiple steps that can be quite complex depending on which operating systems are being used. To make matters worse - some of these machines cannot even access certain files stored between distinct systems. It is typically used in a single network environment with low requirements for performance and low capacity for data exchange. One could say that it is a relatively recent technology implementation application.

SAN (storage area network): 

Micron memory storage is a revolutionary system that utilizes powerful fiber networks to seamlessly connect host servers, allowing for quicker and more secure access to data. SCSI, ESCON, and Fiber Channels are some of the I/O connections used by the system, which are located at the host cluster's back end. For applications with a need for superior network speed, reliability of data, and performance when sharing information, SAN solutions are often the best option due to their superb quality at an elevated price. Take, for instance, the most significant applications of big data in banking and telecom. Leveraging the SCSI block I/O command set, this technology enables rapid random I/O access for faster data transmission through Fiber Channel or disk-level sources. It belongs in high-performance computing because it has high bandwidth and low latency. Nevertheless, SAN systems consisting of 10s of thousands of CPUs are not enough to meet the ever-growing demand and their associated high costs limit their scalability.

Common Devices

Hard Disk Block Storage and disk arrays provide hosts with raw space. On the physical level, block storage provides external services and is pre-formatted with its own file system. Once a system uses it, it becomes monopolized.

You can allocate up to N logical disks across a partition, RAID configuration, or LVM system; for example, when utilizing an array of five disks. However, a physical disk and a logical disk are two distinct concepts. Let's say there are five disks and each disk has 100GB. Additionally, the five logical drives are broken up into five 100 GB-per-drive logical drives. As a result, multiple physical drives form a logical drive.

These logical drives are then mapped by the block storage system to the host. Five hard drives are recognized by the host operating system. The operating system, on the other hand, treats them as just five physical drives because it cannot tell the difference between logical and physical drives. Absolutely not as the operating system would have it. The mounted raw hard disk, which is identical to the built-in hard disk of a typical host, needs to be partitioned and formatted by the operating system in this mode as well.

File Storage 

NAS products typically store files at the file level.

NAS 

Expand your data network by utilizing Network-attached Storage (NAS). NAS are powerful collections of storage devices that grant you direct access to the most important information and services. A cheap system for providing data archiving services is a collection of NAS storage devices. Applications for data storage in the educational, business, and government sectors, for instance.

By leveraging NFS or CIFS command sets and file transfer protocol along with TCP/IP, data access is made smooth, straightforward, and cost-effective. Not only is that but scalability of this system quite remarkable as well. It can be managed without any difficulty due to its user friendly interface. For instance, cluster computing frequently makes use of the NFS file system; however, NAS is unsuitable for high-performance clusters due to its high latency, low bandwidth, and high protocol overhead.

Common Devices

FTP and NFS Servers 

Micron memory storage manufacturers have designed file storage at the file system level to give external services access with a minimum of fuss. All that is necessary for access to this interface is entry into the file system layer.

Using an everyday personal computer, you can easily access and store files with the help of FTP and NFS services. To take advantage of these storage tools, your machine must have the correct operating system and software installed. Then it's simply a matter of connecting to the server shelf after service is completed for seamless use.

Object Storage

High-capacity hard drives connected to a distributed server are the go-to choose for object storage solutions. The object storage software is then installed on several additional servers to serve as management nodes. Other servers can be managed and read-only access is granted by the administration node.

This kind of object storage is used to improve the advantages of file and block storage while also overcoming their drawbacks. To summarize, block storage is slow to read and write, making it unsuitable for sharing.

Can we obtain a storage read/write block that can be shared? 

Object storage comes into play here. A file begins with its properties such as the file's size, modification date, and storage path, among other things, and the contents

FAT32, for instance, is a file system that stores metadata and files directly. The procedure for writing the file to disk begins with splitting it into small chunks using the system's minimum block size, e.g., if a 4K block is required by the file system, then 1000 sections will be created out of that one original document before saving it away.

Metadata and data do not differ in any way. Each block gives you the address of the next block to read, then continues in that order until the file is finished so that you can read all of the blocks. With 100 potential arms to read and write, only one can be used in this scenario due to the impossibility of anticipating where the next block is located without first reading the initial one. This results in a very slow reading and writing speed. 

At this moment, three OSD nodes are transmitting data to the exterior in unison and quickening transmission speed. The speed of reading and writing will get faster if there are more OSD servers. To improve your reading and writing speed, use this method.

On the other hand, the object storage software has a unique file system. This means that the external OSD corresponds to the file server. As a result, it is easy to share files and resolve file-sharing issues. Therefore, the emergence of storage objects combines the advantages of file storage and block storage well.