RustFS

Repository for Laerning Rustfs

RustFS Object Storage

What is RustFS?

RustFS is an open-source object storage server written in Rust. It provides an Amazon S3-compatible API, allowing applications to store and retrieve objects such as documents, images, videos, backups, and datasets using standard S3 tools and SDKs.

How RustFS Works

+------------+      S3 API      +------------------+
| Application| ---------------> |     RustFS       |
+------------+                  +------------------+
                                        |
                           +------------+------------+
                           |                         |
                    Metadata Storage          Object Storage
                           |                         |
                     Object Information      Images, Videos,
                                             Documents, Backups

Key Features

• Written in Rust for safety and performance.
• S3-compatible API.
• Stores unstructured data as objects.
• Scalable and reliable.
• Suitable for cloud-native applications, backups, and data lakes.

Object Storage Concept

Each stored object consists of:

• Object data
• Metadata
• Unique Object ID

Unlike traditional file systems, object storage uses a flat namespace instead of folders, making it highly scalable.

JBOD (Just a Bunch Of Disks)

is a storage configuration where multiple physical disks are used independently rather than combined into a RAID array. In Linux, each disk is treated as its own separate storage device unless you explicitly combine them with tools like LVM or mdadm.

How JBOD Works

Imagine you have four hard drives:

Disk 1  → 500 GB
Disk 2  → 1 TB
Disk 3  → 2 TB
Disk 4  → 4 TB

In a JBOD setup:

• Each disk is visible separately.
• Data is stored on whichever disk you choose.
• There is no striping, no mirroring, and no parity.

Linux

/dev/sda  → 500 GB
/dev/sdb  → 1 TB
/dev/sdc  → 2 TB
/dev/sdd  → 4 TB

You can mount each disk individually:

sudo mount /dev/sdb1 /mnt/disk1
sudo mount /dev/sdc1 /mnt/disk2
sudo mount /dev/sdd1 /mnt/disk3

---

JBOD vs RAID

Feature	JBOD	RAID 0	RAID 1
Data Protection	❌ No	❌ No	✅ Yes
Performance	Normal	High	Read performance can improve
Combines Disks	No (or simple concatenation on some controllers)	Yes	Yes
Disk Sizes	Can differ	Usually matched	Usually matched
Fault Tolerance	None	None	Can survive one disk failure (per mirror)

---

Advantages

• Very simple to set up
• No RAID controller required
• Can mix different disk sizes
• Easy to add or remove disks
• Each disk can use its own filesystem

---

Disadvantages

• No redundancy
• No automatic failover
• If one disk fails, the data on that disk is lost
• No performance improvements from RAID striping

---

Checking Disks in Linux

List available disks:

lsblk

Example:

NAME   SIZE TYPE
sda    500G disk
sdb      1T disk
sdc      2T disk
sdd      4T disk

Show detailed information:

sudo fdisk -l

Or:

sudo parted -l

---

JBOD in Object Storage (e.g., RustFS, MinIO)

Many modern object storage systems, including RustFS and MinIO, often use JBOD-style storage.

Instead of creating a RAID array:

Application
      │
      ▼
   RustFS
      │
 ┌────┼─────┐
 │    │     │
Disk1 Disk2 Disk3 Disk4

The object storage software itself decides:

• where to place objects,
• how to replicate or distribute data,
• how to recover from failures.

This shifts data protection and distribution from the storage hardware (RAID) into the application layer, allowing more flexibility and scalability.

---

When should you use JBOD?

JBOD is a good choice when:

• You want to use disks independently.
• Your application (such as an object storage system) manages data redundancy itself.
• You need flexibility with different disk sizes.
• You don't require RAID-managed redundancy.

For traditional file servers where hardware-level redundancy is important, RAID may be more appropriate. For distributed object storage systems, JBOD is often the preferred approach because the software handles replication and fault tolerance.