클라우드 스토리지 기본 사항: 블록, 파일 및 개체 스토리지

When working with cloud platforms such as AWS, Azure, or Google Cloud, you’ll encounter several different types of storage. The three fundamental categories are:

1. Block storage
2. File storage
3. Object storage

Understanding the difference between them is important when working with VMs, databases, Kubernetes, backups, and cloud architectures.

1. Block Storage

Block storage is the closest equivalent to a physical hard disk or SSD.

The cloud provider gives your VM a virtual disk, and the operating system sees it as a block device.

Cloud
  │
  └── 50 GB Block Disk
          │
          ↓
        VM
          │
       /dev/sdb
          │
       filesystem
          │
        /data

Enter fullscreen mode Exit fullscreen mode

For example, in Linux:

lsblk

Enter fullscreen mode Exit fullscreen mode

might show:

NAME   SIZE
sda     64G
sdb     50G

Enter fullscreen mode Exit fullscreen mode

You can create a filesystem:

sudo mkfs.ext4 /dev/sdb

Enter fullscreen mode Exit fullscreen mode

and mount it:

sudo mkdir /data
sudo mount /dev/sdb /data

Enter fullscreen mode Exit fullscreen mode

Now the disk behaves much like a physical disk attached to the machine.

Examples

Cloud Block storage AWS Amazon EBS Azure Azure Managed Disks Google Cloud Persistent Disk / Hyperdisk

Typical use cases

• VM operating systems
• PostgreSQL/MySQL databases
• Application data
• Kubernetes persistent volumes
• High-performance workloads

2. Object Storage

Object storage works very differently.

Instead of giving you a disk, the cloud provider gives you a storage service accessed through APIs.

The most famous example is Amazon S3.

Application
     │
     │ HTTP / API
     ↓
 Object Storage
     │
     ├── image.jpg
     ├── backup.tar
     ├── data.csv
     └── logs/

Enter fullscreen mode Exit fullscreen mode

You don’t normally get something like:

/dev/sdb

Enter fullscreen mode Exit fullscreen mode

Instead, you interact with objects using APIs, SDKs, CLI tools, or HTTP.

For example, conceptually:

bucket: my-backups
object: database/backup.sql

Enter fullscreen mode Exit fullscreen mode

Examples

Cloud Object storage AWS Amazon S3 Azure Azure Blob Storage Google Cloud Cloud Storage

Typical use cases

• Backups
• Images and videos
• Logs
• Data lakes
• Large files
• Static website assets
• Machine-learning datasets

Why is it popular?

Object storage is designed to scale to enormous amounts of data and is generally very durable.

You don’t have to think about:

/dev/sdb
filesystem
partitions
mount points

Enter fullscreen mode Exit fullscreen mode

You simply store and retrieve objects.

3. File Storage

File storage provides a filesystem over the network.

The cloud provider manages the filesystem, and your machines connect to it.

                 File Storage
                     │
             ┌───────┼───────┐
             │       │       │
            VM1     VM2     VM3
             │       │       │
          /shared  /shared  /shared

Enter fullscreen mode Exit fullscreen mode

Multiple machines can access the same files.

Common protocols include:

• NFS — Network File System
• SMB — Server Message Block

For example, with NFS:

sudo mount -t nfs server:/data /mnt/data

Enter fullscreen mode Exit fullscreen mode

The remote filesystem then appears locally:

ls /mnt/data

Enter fullscreen mode Exit fullscreen mode

Examples

Cloud File storage AWS Amazon EFS / FSx Azure Azure Files / Azure NetApp Files Google Cloud Filestore / NetApp Volumes

Typical use cases

• Shared application files
• Shared configuration
• Multiple VMs accessing the same data
• Kubernetes volumes requiring shared access
• Enterprise applications

4. NFS Is a Protocol, Not a Storage Type

This is an important distinction.

NFS means Network File System.

It is a protocol for accessing a filesystem over a network.

For example:

              NFS
               │
               ↓
        Network filesystem
               │
        ┌──────┼──────┐
        ↓      ↓      ↓
       VM1    VM2    VM3

Enter fullscreen mode Exit fullscreen mode

So don’t think:

NFS = a type of disk.

Instead:

NFS = a way of accessing file storage over a network.

Similarly, SMB is another protocol commonly used for network file storage.

5. Comparing the Three

Block File Object Mental model Virtual disk Network filesystem S3-like bucket Access Block device NFS / SMB API / HTTP Appears as /dev/sdb /mnt/shared Bucket + object key Filesystem You manage it Provider manages it Not a traditional filesystem Mountable Yes Yes Not normally Shared by VMs Limited/depends on service Yes Yes Typical use Databases, VMs Shared files Backups, images, data AWS EBS EFS / FSx S3 Azure Managed Disks Azure Files Blob Storage GCP Persistent Disk Filestore Cloud Storage

6. The Easiest Mental Model

Think about a normal physical computer.

Block storage

You buy a hard drive:

500 GB SSD
     ↓
Computer
     ↓
/dev/sdb
     ↓
ext4
     ↓
/data

Enter fullscreen mode Exit fullscreen mode

Cloud block storage is essentially a virtualized version of this concept.

File storage

Think about a NAS in your office:

                 NAS
                  │
        ┌─────────┼─────────┐
        ↓         ↓         ↓
       PC1       PC2       PC3

Enter fullscreen mode Exit fullscreen mode

Everyone accesses the same shared filesystem.

Cloud file storage provides a similar concept as a managed service.

Object storage

Think about Google Drive/S3-style storage:

Bucket
├── photo.jpg
├── backup.zip
├── video.mp4
└── dataset.csv

Enter fullscreen mode Exit fullscreen mode

Applications access the objects through an API rather than treating the storage as a normal disk.

7. What About SSD vs HDD?

This is a different dimension.

You can have different performance classes within block storage:

Block Storage
    │
    ├── HDD
    ├── Standard SSD
    ├── Premium SSD
    └── High-performance SSD

Enter fullscreen mode Exit fullscreen mode

The important characteristics are typically:

• Capacity
• IOPS
• Throughput
• Latency
• Durability
• Cost

For example, a PostgreSQL database usually cares a lot about IOPS and latency, while a large backup archive may care more about capacity and cost.

8. What About Snapshots and Backups?

Snapshots and backups aren’t really a fourth fundamental storage category.

For example:

Managed Disk
     │
     ├── Snapshot — Monday
     ├── Snapshot — Tuesday
     └── Snapshot — Wednesday

Enter fullscreen mode Exit fullscreen mode

A snapshot is a point-in-time copy/state of storage.

You can often use a snapshot to create a new disk:

Snapshot
    ↓
New Managed Disk
    ↓
Attach to VM
    ↓
Mount filesystem
    ↓
Access files

Enter fullscreen mode Exit fullscreen mode

This is useful when recovering data from a VM.

9. Kubernetes Connection

These concepts become especially useful with Kubernetes.

A Kubernetes application might request storage using a PersistentVolumeClaim:

Pod
 │
 ↓
PersistentVolumeClaim
 │
 ↓
PersistentVolume
 │
 ↓
Cloud Storage

Enter fullscreen mode Exit fullscreen mode

The underlying storage might be:

PVC
 ↓
Azure Managed Disk

Enter fullscreen mode Exit fullscreen mode

or:

PVC
 ↓
Azure Files

Enter fullscreen mode Exit fullscreen mode

The distinction matters because block storage and file storage have different sharing characteristics.

For example, a block-backed volume may typically be attached to one node, while a network file system can allow multiple nodes to access the same filesystem.

10. Summary

The three concepts to remember are:

┌───────────────────────────────────────────┐
│                BLOCK                      │
│                                           │
│ "Give me a disk."                         │
│                                           │
│ /dev/sdb → filesystem → /data             │
└───────────────────────────────────────────┘

┌───────────────────────────────────────────┐
│                 FILE                      │
│                                           │
│ "Give me a shared filesystem."            │
│                                           │
│ NFS / SMB → /mnt/shared                   │
└───────────────────────────────────────────┘

┌───────────────────────────────────────────┐
│                OBJECT                     │
│                                           │
│ "Give me scalable storage through API."   │
│                                           │
│ Bucket → Object → API/HTTP                │
└───────────────────────────────────────────┘

Enter fullscreen mode Exit fullscreen mode

In one sentence:

Block storage is a virtual disk, file storage is a shared network filesystem, and object storage is an API-accessible store for objects such as files and backups.