What is RAID?

In one sentence

RAID is a method of combining multiple physical disks to achieve goals such as increasing performance, increasing failure tolerance, or better storage management.

Definition

If there are several hard drives or SSDs inside a server, they can be placed side by side such that:

• Either they work faster
• Or if one disk fails, data is not lost
• Or both

This general method is called RAID.

RAID stands for Redundant Array of Independent Disks.
RAID is a mechanism at the storage level that converts multiple Physical Disks into one or more Logical Drives / Virtual Disks using techniques like Striping, Mirroring, and Parity.

This can be done in one of the following ways:

• Hardware RAID by a RAID Controller
• Software RAID by the operating system, like mdadm in Linux
• Hybrid / Fake RAID in some systems and motherboards

Why use RAID?

RAID is not just for "combining multiple disks". Its main goal is better storage design.

Benefit	Simple Explanation	Technical Note
Increased Failure Tolerance	If a disk fails, the service may remain active	Depending on the RAID Level, data is recoverable via Mirror or Parity
Performance Improvement	Reading and writing can become faster	Striping and Parallel I/O increase Throughput
Better Use of Multiple Disks	Multiple disks are managed as a single space	Controller or OS converts them into a logical structure
Simplified Management	The OS usually sees one or more logical disks instead of multiple physical ones	RAID presents its output as a Logical Drive

Very Important Note

RAID is not a substitute for Backup.

RAID may protect you against disk failure, but it does not create complete protection against:

• Accidental deletion of files
• File system corruption
• Ransomware
• Human error
• Fire or total loss of the server

So even if you have RAID, you still need a real Backup.

How does RAID work?

In its simplest form, multiple physical disks enter a RAID layer, and its output is displayed as a logical disk to the operating system.

In this path:

1. Physical Disks are placed inside the server
2. RAID Controller or Software RAID manages them
3. The output of this layer is usually a Logical Drive
4. The Operating System sees this output as a normal disk
5. Then Partition, LVM, or File System is built on it

Basic Concepts in RAID

To understand RAID, you need to know three main concepts:

1) Striping

In Striping, data is distributed across multiple disks.
This causes multiple disks to work on data simultaneously, improving performance.

• Main advantage: Performance
• Main limitation: Usually does not create Fault Tolerance on its own

2) Mirroring

In Mirroring, an identical copy of data is written to another disk.

• Main advantage: Failure tolerance
• Main limitation: Part of the capacity is consumed for data copying

3) Parity

Parity is extra information that helps RAID rebuild data if one or more disks fail.

• Main advantage: Balance between capacity and failure tolerance
• Main limitation: Parity calculation affects write performance

Parity Simply Put

Parity can be considered a type of "auxiliary information for data reconstruction".
With the help of this information, RAID in some Levels can recalculate and recover data after a disk failure.

Common RAID Levels

Not all RAIDs are the same. Each RAID Level is suitable for a specific need.

Quick Comparison

RAID Level	Minimum Disks	Usable Capacity	Failure Tolerance	Performance	Common Usage
RAID 0	2	Sum of all disks	❌ None	Very high	Temp data, Cache, Non-critical Workload
RAID 1	2	Approx 50%	✅ Failure of 1 disk per Mirror	Good read, Normal write	Operating System, Boot Volume
RAID 5	3	Total disks minus 1 disk	✅ Failure of 1 disk	Good read, Moderate write	File Server, Capacity-focused
RAID 6	4	Total disks minus 2 disks	✅ Failure of 2 disks	Good read, Weaker write than RAID 5	Archive, Capacity with more safety
RAID 10	4	Approx 50%	✅ Min 1 disk and in some layouts more	Very good	VM, Database, High-traffic Workload

Each RAID Level Simply Explained

RAID 0

In RAID 0, data is split across multiple disks, but no copy or Parity exists.

• Advantage: Very fast
• Weakness: If only one disk fails, the entire Array is lost

Risk of RAID 0

RAID 0 is only suitable for scenarios where data loss is acceptable or data is kept elsewhere.

RAID 1

In RAID 1, data is written to two disks in a mirrored fashion.

• Advantage: Failure of one disk usually does not cause service loss
• Weakness: Half of the capacity is spent on Mirror

Common uses:

• OS disk
• Boot Volume
• Small servers focusing on simplicity and stability

RAID 5

In RAID 5, data and Parity are distributed across disks.

• Advantage: Usable capacity better than RAID 1 and RAID 10
• Weakness: In write operations, Parity calculations reduce performance
• Failure Tolerance: Failure of 1 disk

Common uses:

• File Server
• Workloads with more reading than writing

RAID 6

RAID 6 is similar to RAID 5 but uses two Parities instead of one.

• Advantage: Tolerance for failure of 2 disks
• Weakness: Slower writing than RAID 5
• Usage: Archive, large volumes, where long Rebuild risk matters

RAID 10

RAID 10 is a combination of Mirroring + Striping.

• Advantage: High Performance + Suitable Fault Tolerance
• Weakness: Only about 50% usable capacity
• Usage: Virtualization infrastructure, Databases, VM Storage

Popular Choice in Infrastructure

In many professional infrastructures, RAID 10 is one of the best options for sensitive and high-traffic Workloads because it behaves better than RAID 5/6 in writing and Rebuild.

Example with 4 x 1TB Disks

Assume you have 4 disks of 1TB.

RAID Level	Usable Capacity	Failure Tolerance
RAID 0	4TB	0 disks
RAID 1	Usually 2TB if two mirror pairs	Failure of 1 disk per Mirror pair
RAID 5	3TB	1 disk
RAID 6	2TB	2 disks
RAID 10	2TB	Failure of 1 disk per Mirror pair

Caution in Interpreting RAID 10

In RAID 10, you cannot simply say "always 2 disks can fail".
If both failed disks are in the same Mirror pair, the entire Array is lost.
So RAID 10's failure tolerance also depends on the pattern of disk failures.

Hardware RAID and Software RAID

RAID can be implemented in two main ways.

Comparison

Feature	Hardware RAID	Software RAID
Implementation Location	RAID Controller	Operating System
Management	Via BIOS/Controller Utility	With tools like mdadm
Dependency on OS	Less	More
Boot Capability	Usually simpler	Depends on configuration
Dedicated Cache	Usually has it	Usually does not have it
Cost	More	Less
Flexibility	Vendor dependent	Very high in Linux

When to use Hardware RAID?

• Enterprise Servers
• When there is a professional controller with Cache and Battery/FBWC
• When you want the OS to see only the final output of RAID

When to use Software RAID?

• Linux servers without a professional RAID Controller
• Environments where hardware simplicity or Vendor independence is important
• Scenarios where management via standard Linux tools is preferred

Document Focus

In many organizational scenarios, when we talk about RAID at the server layer, we usually mean Hardware RAID, unless explicitly mentioning Software RAID.

What is the relationship between RAID and Logical Drive?

One of the most important points is that RAID itself is not the disk that the OS sees.
The output of RAID is usually a Logical Drive.

Clarification

RAID is a method for organizing and protecting multiple disks, but Logical Drive is its final output for the OS.

• RAID = Mechanism or Structure
• Logical Drive = The disk the OS observes

Important Operational Concepts in RAID

Rebuild

When a disk fails and is replaced by a new disk, RAID starts rebuilding data onto the new disk. This process is called Rebuild.

• Performance usually decreases during Rebuild
• In large Arrays, Rebuild may be time-consuming
• During this period, the risk of a second failure becomes more important

Hot Spare

A Hot Spare is a disk present in the system as ready but inactive.
If one of the Array's disks fails, RAID can automatically start Rebuilding onto the Hot Spare.

Cache

Some RAID Controllers have Write Cache / Read Cache.
This Cache can greatly improve performance, but its safety status must also be checked.

Cache Safety

If Write Cache is active but proper protection like Battery Backup or Flash-Backed Write Cache does not exist, data may be lost during a power outage.

How to choose the right RAID?

Choosing RAID depends on these questions:

Question	Effect on RAID Selection
Is performance or capacity more important?	If performance is important, RAID 10 is usually better
Do we have heavy writing or heavy reading?	RAID 5/6 are weaker in writing than RAID 10
Should failure of two disks be tolerated?	RAID 6 is more suitable than RAID 5
Is cost important?	RAID 5/6 provide better capacity than RAID 10
Is long Rebuild worrying?	On large disks, RAID 10 or RAID 6 is usually more logical than RAID 5

Quick Selection Summary

Scenario	Recommended RAID
Boot / OS	RAID 1
VM Storage / Database	RAID 10
Archive / Capacity-Focused Storage	RAID 6
Read-Heavy General File Storage	RAID 5 or RAID 6
Temp / Scratch Data	RAID 0 only if risk is acceptable

Role of RAID in Plaivid

In the Plaivid architecture, RAID is part of correct storage layer design, especially when using Local Storage.

The common pattern could be as follows:

Part	Common RAID	Reason
OS	RAID 1	Simplicity, Stability, Safer Boot
VAR / Log	RAID 1 or part of System Storage	Separation of variable data
DATA / VM Workloads	RAID 10	Better performance and better behavior in Writes
Local Backup	RAID 6 or RAID 10	Depending on priority of capacity or performance

Design Note

If Plaivid runs on Shared Storage like SAN, NAS, or Distributed Storage, RAID selection might have been done at a lower layer outside the server itself.

Common Myths About RAID

1) RAID means Backup

No. RAID is only part of High Availability and Fault Tolerance, not a replacement for Backup.

2) The more complex the RAID, the better

No. A suitable RAID must align with Workload, Number of Disks, Budget, and Design Goal.

3) Multiple Logical Drives on one Array means completely separate performance

No. If multiple Logical Drives are created from the same set of disks, they still use the same shared I/O behind the scenes.

4) RAID 5 is always an economical and good choice

Not necessarily. On large disks and heavy workloads, Rebuild time and error risk can turn RAID 5 into a weaker option.

Important Notes for IT Experts

• Use disks of the same size and type; usually, the smallest disk becomes the basis for capacity
• Manage Controller Firmware and Disk Firmware carefully
• Monitor status of Predictive Failure, Media Error, and Battery/Cache
• On large Arrays, take URE and Rebuild time seriously
• For heavy Workloads, check Stripe Size and File System Alignment
• Before Production, test scenarios of Disk Failure / Rebuild / Controller Failure
• If using Software Raid, activate monitoring tools like mdadm --detail and Alerting

Best Practice

For virtualization infrastructures and sensitive Workloads, besides choosing the right RAID, planning for Monitoring, Spare Disk, Backup, and Recovery Testing is also necessary.

Conclusion

RAID is a method for combining multiple disks to strike a balance between performance, capacity, and failure tolerance.

To put it very simply:

• RAID 0 = Fast, but without protection
• RAID 1 = Simple and reliable for the operating system
• RAID 5 = Better capacity, tolerance for 1 disk failure
• RAID 6 = Safer for large capacities
• RAID 10 = Powerful choice for VMs and high-traffic Workloads

And most importantly:

• RAID is different from Logical Drive
• RAID is different from Backup
• Choosing RAID must be based on the real needs of the infrastructure, not just raw capacity