Backup Storage: NAS, Cloud Object Storage, and 3-2-1 Backup Strategy

Building backup storage is one of the important things to consider when developing an application, especially an application that is already running in a production environment. Why? The goal is simple: to ensure that data can still be recovered when the primary server experiences failure, disruption, or data loss.

Imagine an application that is already being used by many users, and suddenly the production database experiences a problem or some of its data becomes corrupted. Damn...! The application may start throwing errors, services may become unavailable, and in certain situations, we could even lose data. In addition, the recovery process may take a significant amount of time if we do not have a proper backup mechanism.

Therefore, backup storage becomes an important component of a production application architecture. A backup provides a copy of data that can be used for recovery when something goes wrong with the primary database or storage.

There are various approaches that can be used to store backups. In this article, we will discuss two storage options that can be used in a backup architecture: Network Attached Storage (NAS) and Cloud Object Storage. We will also discuss how the 3-2-1 backup strategy can be applied to reduce the risk of data loss.

What Is Network Attached Storage (NAS)?

Understanding NAS

NAS (Network Attached Storage) is a storage server or appliance connected to a network. NAS can be used to store:

• Database backups.
• Shared documents and files.
• Snapshots.
• Application archives.

NAS is generally designed as centralized storage with capacity that can be adjusted based on your needs. It can be accessed by multiple servers or devices over the network according to the permissions granted to them. NAS is suitable when an organization wants to manage its own storage and have greater control over its data and infrastructure.

Network File System (NFS) Inside NAS

NFS (Network File System) is a protocol that allows a server to access files or directories on another machine over a network.

NFS is not a storage device. NAS is the storage server, while NFS is one of the protocols that can be used to access files stored on a NAS.

NAS         = storage server
NFS         = network file access protocol
Mount point = local directory tempat NFS share di-mount

NAS         = storage server
NFS         = network file access protocol
Mount point = local directory tempat NFS share di-mount

For example, a NAS provides an NFS share for storing backups. The application server then mounts the NFS share to a local directory such as /mnt/backup-nas.

From the application server's perspective, this directory can be used like a regular local directory, but the data written to it is actually stored on the NAS over the network.

Example of Backing Up Data to NAS

First, we need to configure the backup tool. A backup tool is software, a script, or a service responsible for creating and managing the backup process.

In this example, the backup tool first creates a database backup on local storage. Once the backup process is complete, the backup file is copied to the NFS mount point, which stores the file on the NAS.

Example directory configuration:

Local backup: /var/lib/backup-tool/backups
NFS mount:    /mnt/backup-nas
Destination:  /mnt/backup-nas/application/backups

Local backup: /var/lib/backup-tool/backups
NFS mount:    /mnt/backup-nas
Destination:  /mnt/backup-nas/application/backups

The backup flow looks like this:

Production Database
        │
        ▼
   Backup Tool
(e.g., backup script)
        │
        ▼
   Local Storage
        │
        │ Copy / Sync
        ▼
   NFS Mount Point
        │
        │ NFS
        ▼
       NAS

Production Database
        │
        ▼
   Backup Tool
(e.g., backup script)
        │
        ▼
   Local Storage
        │
        │ Copy / Sync
        ▼
   NFS Mount Point
        │
        │ NFS
        ▼
       NAS

For example, a backup script can execute a utility such as pg_dump for PostgreSQL, mysqldump for MySQL, or another database-specific backup command. The resulting backup file is temporarily stored on local storage before being copied to the NAS.

Each application should have a dedicated backup directory on the NAS with appropriate permissions. Before performing write or delete operations, the backup process should also validate the destination path to reduce the risk of accidentally modifying or deleting data belonging to another application's directory.

Cloud Object Storage

What Is Cloud Object Storage?

Cloud Object Storage is a cloud-based storage service that stores data as objects inside buckets. This type of storage is commonly used to store database backups, files, logs, archives, and other types of unstructured data.

A bucket is the main container used to store objects in object storage.

A simple way to imagine it:

Object Storage = warehouse complex
Bucket         = one warehouse
Object         = item inside the warehouse

Object Storage = warehouse complex
Bucket         = one warehouse
Object         = item inside the warehouse

Types of Cloud Storage

Cloud storage is a general term for data storage services provided by cloud providers. In general, cloud storage can be categorized into Object Storage, File Storage, and Block Storage.

• Object Storage stores data as objects that are generally accessed through an API. This type of storage is commonly used for backups and archives because it can be easily scaled. Examples include Amazon S3, Azure Blob Storage, and Google Cloud Storage.
• File Storage stores data using a file and directory structure. It is generally accessed through protocols such as NFS or SMB (Server Message Block) and can be mounted on a server. Examples include Amazon EFS, Azure Files, and Google Filestore.
• Block Storage provides storage in the form of blocks that are recognized by the server like a disk. It can be formatted using a filesystem such as ext4 or XFS. Examples include Amazon EBS, Azure Managed Disks, and Google Persistent Disk.

Simply put, the three can be distinguished as follows:

Cloud Storage
│
├── Object Storage
│   └── Bucket + Object
│       Access: API / SDK
│
├── File Storage
│   └── File + Directory
│       Access: NFS / SMB
│
└── Block Storage
    └── Virtual Disk / Block Device
        Access: attached to server

Cloud Storage
│
├── Object Storage
│   └── Bucket + Object
│       Access: API / SDK
│
├── File Storage
│   └── File + Directory
│       Access: NFS / SMB
│
└── Block Storage
    └── Virtual Disk / Block Device
        Access: attached to server

The easiest way to remember them:

Object Storage → "I have a bucket"
File Storage   → "I have a shared folder"
Block Storage  → "I have a disk"

Object Storage → "I have a bucket"
File Storage   → "I have a shared folder"
Block Storage  → "I have a disk"

How Backups Work with Cloud Object Storage

With Cloud Object Storage, backups do not need to be accessed through an NFS mount like in the previous NAS example. A backup tool can upload backup files to a bucket using an API, SDK, or CLI provided by the storage provider.

One example of the backup flow is:

Production Database
        │
        ▼
   Backup Tool
(e.g., backup script)
        │
        ▼
   Local Storage
        │
        │ Upload via API / SDK
        ▼
Cloud Object Storage
        │
        ▼
      Bucket
        │
        └── database-backup.sql

Production Database
        │
        ▼
   Backup Tool
(e.g., backup script)
        │
        ▼
   Local Storage
        │
        │ Upload via API / SDK
        ▼
Cloud Object Storage
        │
        ▼
      Bucket
        │
        └── database-backup.sql

For example, the backup tool generates a backup file:

/var/lib/backup-tool/backups/database-backup.sql

/var/lib/backup-tool/backups/database-backup.sql

The file is then uploaded to a bucket:

production-backups/
└── database/
    └── database-backup.sql

production-backups/
└── database/
    └── database-backup.sql

The main difference compared to the previous NAS example is how the storage is accessed.

NAS
Backup → Local Storage → NFS Mount → NAS

Cloud Object Storage
Backup → Local Storage → API / SDK → Bucket

NAS
Backup → Local Storage → NFS Mount → NAS

Cloud Object Storage
Backup → Local Storage → API / SDK → Bucket

With NAS, the server can access remote storage as a filesystem through a protocol such as NFS. Meanwhile, with Cloud Object Storage, an application or backup tool generally interacts with the storage through an API or SDK to perform operations such as uploading, downloading, listing, and deleting objects.

The 3-2-1 Backup Strategy

The 3-2-1 backup strategy is an approach used to reduce the risk of data loss by keeping multiple copies of data across different storage systems and locations.

The principle is quite simple:

3 → Keep at least three copies of the data
2 → Store the copies on two different media or storage systems
1 → Keep at least one backup in an offsite location

3 → Keep at least three copies of the data
2 → Store the copies on two different media or storage systems
1 → Keep at least one backup in an offsite location

With this strategy, a failure of one server or storage system does not immediately cause all copies of the data to be lost.

Three Copies of Data

The first principle is to have at least three copies of the data, consisting of one primary copy and at least two backup copies.

Simply put:

1 Production Copy + 2 Backup Copies = 3 Copies

1 Production Copy + 2 Backup Copies = 3 Copies

For example:

1. Production Database
2. Backup Copy #1
3. Backup Copy #2

1. Production Database
2. Backup Copy #1
3. Backup Copy #2

Two Different Media or Storage Systems

Backup copies should not all depend on the same disk, server, or storage system.

Several possible storage combinations include:

Option A
Production Database
├── Local Backup
└── NAS

Option B
Production Database
├── Local Backup
└── Cloud Object Storage

Option C
Production Database
├── NAS
└── Cloud Object Storage

Option A
Production Database
├── Local Backup
└── NAS

Option B
Production Database
├── Local Backup
└── Cloud Object Storage

Option C
Production Database
├── NAS
└── Cloud Object Storage

The main idea is to reduce the risk of a single point of failure. If one storage system fails, another backup copy is still available on a different storage system.

One Offsite Copy

The final principle is to keep at least one backup in an offsite location.

Offsite means that the backup is stored in a location or failure domain that is separate from the primary infrastructure. This helps prevent a failure at the primary location from affecting all backup copies at the same time.

Both NAS and Cloud Object Storage can be used as offsite backup options:

NAS
└── Offsite if located at another site / data center

Cloud Object Storage
└── Can be used as offsite storage
    if separated from the primary failure domain

NAS
└── Offsite if located at another site / data center

Cloud Object Storage
└── Can be used as offsite storage
    if separated from the primary failure domain

Conclusion

Backup is not only about creating copies of data. It is also about how and where those copies are stored so they remain available when the primary system fails.

NAS can be a good option when centralized storage needs to be self-managed and accessed over a network, for example through NFS. Meanwhile, Cloud Object Storage provides an object-and-bucket-based approach that is generally accessed through APIs or SDKs and can be used as an option for storing backups outside the primary infrastructure.

In practice, NAS and Cloud Object Storage do not have to be used together. They are storage options that can be selected or combined depending on the organization's needs, infrastructure, costs, and policies.

More importantly, we need to think about how the backup architecture itself is designed. The 3-2-1 backup strategy can be used as a baseline: keep at least three copies of the data, store them across two different media or storage systems, and keep at least one backup in an offsite location.

References

• PostgreSQL backup best practices — 15 essential PostgreSQL backup strategies for production systems
• What Is Cloud Storage? And What Are the Different Types of Cloud Storage?
• Boost Your Data Security with the 3–2–1 Backup Strategy: Safeguarding Your Vital Information