Skip to main content
StashRaid

Storage Parameters

Usable Capacity

12 TB

Fault Tolerance

4

Array Resilience

Architecture

RAID 5 • Parity

Fault Tolerance

1 Concurrent Failures Allowed

Current Configuration

RAID 5 (Parity)

12 TB Usable

Total RAW: 16 TB

Disk 01DATA
Disk 02DATA
Disk 03DATA
Disk 04PARITY

Total Drives

4 x 4 TB

Efficiency

75% Utilized

Usable: 12 TBWasted to redundancy: 4 TB

Pros & Cons

  • PRO: Good balance of capacity/speed/safety

  • PRO: Cost-effective redundancy

  • CON: Slower writes due to parity calculation

    This setup prioritizes redundancy but has associated performance or capacity costs.

STORAGE ARCHITECTURE // COMPARATIVE ANALYSIS

RAID vs. JBODThe Complete Guide.

Understanding the trade-offs between RAID levels and the three different things people call "JBOD" — spanning, pass-through, and enclosure semantics. When to stripe, when to span, and when to Just Use A Bunch Of Disks.

What JBOD Actually Means

Just a Bunch Of Disks — Three Meanings, One Acronym

The term "JBOD" is notoriously overloaded. Depending on who you ask — a hardware vendor, a storage administrator, or a home-lab enthusiast — it can mean three fundamentally different things. Understanding which "JBOD" someone is referring to is half the battle.

1. Enclosure Pass-Through

A hardware enclosure (e.g. a disk shelf with a SAS expander backplane) that simply exposes each physical drive as an independent device to the host controller. No RAID firmware, no virtualisation — the OS sees four individual disks. This is the purest definition: a box full of drives, no abstraction layer.

2. Concatenation / Spanning

A logical volume manager (LVM, mdadm, Storage Spaces) combines multiple disks into one large span. Data fills disk 1 first, then spills to disk 2, then disk 3. This is what most consumer NAS interfaces call "JBOD" — a single logical volume built from heterogeneous drives with no striping and no parity.

3. Separate Volumes

The RAID controller or OS creates N separate logical volumes — one per physical drive — rather than aggregating them. Each disk appears as its own drive letter or mount point (D:, E:, F:, G:). This is common in older external multi-bay enclosures where the controller has a "JBOD" mode that just labels each disk.

Watch out: A vendor claiming "JBOD support" could mean any of these three. Always check whether they mean spanning, pass-through, or independent volumes — they behave completely differently under failure.

RAID 0 vs. JBOD: Performance, Capacity & Failure

Striping vs. Spanning vs. Pass-Through — Three Different Animals

RAID 0 — Striped for Speed

RAID 0 splits data into blocks (typically 64 KB or 128 KB) and writes them across all drives in parallel. With 4 drives, a 1 MB file gets written in ~256 KB chunks simultaneously — delivering nearly 4× the sequential throughput of a single disk.

Read Speed~4× Single Disk
Write Speed~4× Single Disk
Failure ToleranceZero — Total Loss
Storage Layout — 4 × 4 TB Drives
HDD 1
A1
B1
C1
D1
HDD 2
A1
B1
C1
D1
HDD 3
A1
B1
C1
D1
HDD 4
A1
B1
C1
D1
Data striped across all drives

Failure Behaviour

What Happens When a Drive Dies

RAID 0 — Total Loss

A single drive failure destroys the entire array. Because data is interleaved across all disks, there are no complete files on any individual drive. Recovery is effectively impossible without professional cleanroom services, and even then success is not guaranteed. RAID 0 is gambling your data for speed.

JBOD Spanned — Partial Loss

Only files physically stored on the failed drive are lost. Files on the surviving disks remain intact and accessible. The logical volume is broken — you lose access to the span as a whole — but the surviving drives can be imported individually to recover their data.

JBOD Individual — Isolated Loss

Only the failed drive's volume is lost. Every other drive continues operating independently. This is the best failure isolation model of the non-redundant options. Combined with per-drive backups or a parity model (unRAID), this becomes a highly resilient setup.

RAID 5 — Survives One Failure

The array stays online and readable after a single drive failure. Parity data on the remaining drives reconstructs missing blocks on the fly. However, a second failure during rebuild (the "RAID 5 write hole") means total loss. RAID 5 rebuild times on large drives can take days — a real risk window.

Use Cases

When to Pick Each Configuration

RAID 0 — Scratch / Cache / Temp

Any workload where speed matters more than durability: video editing proxies, game install caches, temporary build artifacts, Photoshop scratch disks. The data is either disposable or backed up elsewhere.

No redundancy — assume total loss risk
JBOD Spanned — Media Servers

Perfect for Plex, Jellyfin, Emby or any media-collection server where (a) you mix drive sizes, (b) you don't want to waste capacity on parity, and (c) losing a single drive means losing only that drive's media (not the whole library). This is the unRAID model.

Great for large, slow-growing collections
JBOD Pass-Through — ZFS

ZFS wants raw disk access. A JBOD enclosure or HBA in IT mode passes each drive as an independent device to the OS, where ZFS builds its own vdevs (mirror, RAID-Z, etc.). This gives ZFS full control over data layout, checksumming, and snapshots — the gold standard for data integrity.

Best practice for ZFS / TrueNAS / Proxmox

When JBOD Saved My Data

Real-World Recovery Stories

The most compelling argument for JBOD (spanning mode, specifically) comes from the hard-learned lessons of home-lab and media-server operators. Here is a scenario that plays out repeatedly:

Story: The 4-Drive RAID 0 That Died

A home-lab operator runs an 8 TB RAID 0 array (4 × 2 TB SSDs) for a game library. One drive develops bad sectors. The array is dead. 8 TB of Steam games, gone. Recovery quote: $2,400+.

Compare that to the same 8 TB in a spanned JBOD volume on unRAID: one drive fails, only the files physically on that 2 TB drive are lost — and if parity was enabled, zero data loss. The remaining 6 TB stays online and accessible during replacement.

This is the unRAID model in action: JBOD spanning for mixed-size flexibility, plus single or dual parity for protection. It is slower than RAID on paper, but for media and bulk storage the throughput difference is transparent to a Plex stream or file copy — while the resilience difference is everything.

Story: The Failed HBA in a JBOD Shelf

A homelabber's LSI 9200 HBA fails in a SuperMicro JBOD enclosure. Because the enclosure passes drives individually (no RAID firmware), every disk is still readable. They plug the drives into a new HBA, ZFS imports the pool, and the array is back online in 30 minutes.

If this had been a hardware RAID controller with a foreign metadata stripe, recovery would have required an identical controller model with matching firmware — a much harder and more expensive problem. JBOD pass-through saved the data simply by staying dumb.

Capacity Comparison: 4 × 4 TB Drives

Raw Capacity vs. Usable Capacity vs. Parity Overhead

RAID 0
Raw

16 TB

Usable

16 TB (100%)

Redundancy

None — total loss

Best For

Scratch / Temp / Cache

JBOD Spanned
Raw

16 TB

Usable

16 TB (100%)

Redundancy

None — partial loss

Best For

Media servers / unRAID

JBOD Individual
Raw

4 × 4TB

Usable

16 TB (100%)

Redundancy

None — isolated loss

Best For

ZFS / TrueNAS

RAID 5
Raw

16 TB

Usable

12 TB (75%)

Redundancy

1-drive fault toler.

Best For

General file server

RAID 10
Raw

16 TB

Usable

8 TB (50%)

Redundancy

1-2 drives (mirror)

Best For

Databases / High IOPS

Key insight: RAID 0 and JBOD spanning give you the same usable capacity (16 TB from 4×4 TB), but with critically different failure profiles. RAID 5 costs 25% of raw capacity for single-drive redundancy. RAID 10 costs 50% but offers faster rebuilds and better IOPS. There is no one-size-fits-all answer — the right choice depends entirely on your workload's speed, capacity, and resilience requirements.

Frequently Asked Questions

Quick Answers to Common Questions

STORAGE_ARCHIVE_V1.0//RAID vs JBOD Comparative Guide
REFERENCE DOCUMENT — NOT A BACKUP STRATEGY