Redundant Array of Independent Disks (RAID) is a data storage virtualisation technology that combines multiple physical disk drive components into a single logical unit for the purposes of data redundancy, performance improvement, or preferably both. For our customers running bare metal servers, understanding RAID is crucial for data integrity and system performance.
In this guide, Daniel Paulini, our Server Delivery Manager, outlines the basic RAID levels, the difference between hardware and software RAID, and Novoserve's policies regarding advanced RAID configurations.
The Basic RAID Levels
Different RAID levels offer varying balances of performance, redundancy, and storage capacity. Below are the most common levels our customers utilize:
RAID 0 (Striping)
RAID 0 is designed purely for speed and performance. Data is split into blocks and written across all drives in the array. While offering the best read and write performance, Novoserve does NOT recommend this option for any important or mission critical data which is not backed up in a different location.
|
Advantage |
Disadvantage |
|---|---|
|
Excellent read and write performance |
No data redundancy |
|
Maximum usable storage capacity (N drives) |
If one drive fails, all data is lost |
|
Simple implementation |
Not suitable for mission-critical data |
RAID 1 (Mirroring)
RAID 1 provides full data redundancy by mirroring the data exactly between two or more drives. If one drive fails, the other(s) take over.
|
Advantage |
Disadvantage |
|---|---|
|
High fault tolerance (good data protection) |
Lowest usable storage capacity (only 50% of total) |
|
Excellent read performance |
Slower write performance compared to RAID 0 |
|
Simple recovery process |
More expensive per gigabyte of usable storage |
RAID 5 (Striping with Distributed Parity)
RAID 5 is one of the most common configurations, balancing performance and fault tolerance. Data is striped across all drives, and a block of parity data (used for recovery) is distributed across the drives. It requires a minimum of three drives.
|
Advantage |
Disadvantage |
|---|---|
|
Good read performance |
Write performance can be slow |
|
High usable storage capacity (N-1 drives) |
Can only tolerate a single drive failure |
|
Efficient use of disk space |
Longer rebuild times after a failure, increasing risk of a second failure |
RAID 6 (Striping with Dual Parity)
RAID 6 is an extension of RAID 5, adding a second independent parity block distributed across the drives. This allows the array to withstand the simultaneous failure of two drives. It requires a minimum of four drives.
|
Advantage |
Disadvantage |
|---|---|
|
Very high fault tolerance (can tolerate two drive failures) |
Slower write performance than RAID 5 due to calculating two parity blocks |
|
Good read performance |
Longer rebuild times |
|
High usable storage capacity (N-2 drives) |
More drives required than RAID 5 |
RAID 10 (or 1+0) (Striping and Mirroring)
RAID 10 combines the benefits of RAID 1 (mirroring) and RAID 0 (striping). Data is first mirrored, and then the mirrors are striped. This configuration offers both high performance and excellent fault tolerance. It requires a minimum of four drives (or any larger even number).
|
Advantage |
Disadvantage |
|---|---|
|
Excellent read and write performance |
Expensive (only 50% usable capacity) |
|
Very high fault tolerance |
Complex setup and management compared to simpler levels |
|
Fast rebuild times |
Minimum of four drives required |
Hardware RAID vs. Software RAID
Novoserve offers bare metal servers from HPE and Supermicro, some of which may be equipped with dedicated Hardware RAID (HW RAID) controllers. It is important for our customers to understand the difference between the two main types of RAID.
Hardware RAID (HW RAID)
HW RAID utilizes a dedicated controller card (or integrated chip on the motherboard) with its own processor and cache memory to manage the RAID array.
|
Characteristic |
Description |
|---|---|
|
Performance |
Generally higher performance, as it offloads the RAID calculation tasks from the main CPU. |
|
Setup |
Managed via the controller's BIOS/firmware interface (pre-boot environment). |
|
Portability |
Limited portability. If the controller fails, it often needs to be replaced with an identical model to preserve the array configuration and access data. |
|
Cost |
Adds to the cost of the server due to the specialized hardware. |
Software RAID (SW RAID)
SW RAID manages the array using the operating system's kernel and the server's main CPU and memory.
|
Characteristic |
Description |
|---|---|
|
Performance |
Relies on the main server CPU, which can introduce a slight performance overhead. Modern CPUs make this impact minimal for most workloads. |
|
Setup |
Configured directly within the operating system (e.g., via mdadm on Linux or Storage Spaces on Windows Server). |
|
Portability |
Highly portable. The RAID configuration data is stored directly on the drives themselves. If a server component fails (even the motherboard), the drives can be moved to another compatible machine, and the OS can typically re-import the array. |
|
Cost |
No additional hardware cost; relies on existing system resources. |
While we provide servers with both options, Novoserve prefers and recommends Software RAID for most customer deployments.
This preference is due to the enhanced data portability and resilience offered by SW RAID. Since the RAID configuration data is stored directly on the drives and not on a proprietary controller, it simplifies the recovery process in the event of a hardware failure. If a hardware RAID card fails, replacing it with an identical, or at least compatible, model is necessary to read the array. With SW RAID, the drives can be easily migrated to another bare metal server, and the operating system can rebuild the array, ensuring faster and more predictable recovery times.
Advanced Configurations and Customer Flexibility
As a bare metal server provider, Novoserve prioritizes giving our customers maximum control over their infrastructure.
ZFS, BTRFS, and JBOD
While Novoserve does not offer pre-installation or support for setting up advanced volume managers like ZFS (Zettabyte File System) or BTRFS, or non-RAID configurations like JBOD (Just a Bunch of Disks), our bare metal environment means you have complete administrative freedom.
Our customers are completely free to install, configure, and manage any compatible operating system and storage configuration they desire, including ZFS, BTRFS, or JBOD setups.
However, it is important to note that any configuration outside of standard OS-level Software RAID (like Linux mdadm or Windows Storage Spaces) falls under the customer's full responsibility for setup, monitoring, and maintenance.
Further Information
We recommend keeping following best practices in mind:
Understanding Rebuilds: Be aware that rebuilding a degraded RAID array (especially RAID 5 or 6) is a high-stress operation. During the rebuild, performance will be reduced, and the risk of a second disk failure increases.
Hot Spares: Consider configuring a spare disk (Hot Spare) in your RAID array. This drive remains idle but immediately begins replacing a failed drive automatically, minimizing downtime and data vulnerability.
Monitoring is Key: Regardless of the RAID level or type (HW/SW), consistent monitoring of drive health is non-negotiable. Utilize SMART data monitoring tools or the RAID utility provided by your OS or controller to detect pre-failure indicators.
RAID is Not a Backup: RAID is a solution for high availability and fault tolerance, preventing downtime from a disk failure. It is not a substitute for a comprehensive backup strategy. RAID cannot protect against accidental deletion, file corruption, or malware, as the corrupted data will simply be mirrored across the array.