In enterprise server infrastructures, data security and disk performance are the cornerstones of business continuity. PERC (PowerEdge RAID Controller) hardware on Dell PowerEdge servers allows you to organize your disks in a redundant and high-performance manner. However, an incorrectly configured RAID architecture can lead to performance bottlenecks and data loss risks, no matter how fast your disks are.
In this guide, we will discuss the best RAID configuration practices, PERC settings, and performance optimizations on Dell PowerEdge servers.
Choosing the Right RAID Level
The first and most important step of RAID configuration is choosing the RAID level that best fits your workload. Each RAID level offers different balances between performance, capacity, and redundancy.
1. RAID 10: Critical Databases and High IOPS
If you are running Microsoft SQL Server, Oracle, or write-heavy virtualization hosts on your server, RAID 10 is the level you should choose.
- Advantage: Offers the highest read/write performance and redundancy by using both striping and mirroring.
- Disadvantage: 50% of disk capacity is allocated for redundancy.
2. RAID 5 and RAID 6: File Servers and General Storage
For workloads that are not write-heavy, are mostly read-oriented, or require large storage capacity, RAID 5 or RAID 6 is appropriate.
- RAID 5: Built with at least 3 disks, can tolerate the failure of 1 disk.
- RAID 6: Built with at least 4 disks, can tolerate the simultaneous failure of 2 disks using double parity. RAID 6 should be preferred in large capacity SATA/SAS disk groups.
PERC Controller Settings and Optimization
When creating a RAID group on Dell PowerEdge servers, correctly configuring some critical parameters in the PERC interface (iDRAC or Lifecycle Controller) directly impacts your performance.
Stripe Element Size
Stripe size is the block size into which data is divided when written onto disks.
- For Databases and Virtualization: A stripe size of 64 KB or 128 KB is ideal.
- For Large File Storage and Streaming: A stripe size of 256 KB or 512 KB should be preferred.
Read Policy
- Read Ahead: Allows the controller to predict sequential data and cache it. This significantly increases performance in sequential read workloads.
- No Read Ahead: Can be preferred in random read database workloads.
Write Policy
- Write Back: Data is written directly to the fast cache on the PERC, and the controller signals the operating system that the operation is complete. This setting can increase write performance by 5 to 10 times.
- Write Through: The operation is not considered complete until data is written directly to the disk. It is safe but very slow.
Important Note: To enable the Write Back policy, the battery backup unit (BBU) or capacitor (CacheVault) running on your PERC card must be in a healthy state. Otherwise, cache data may be lost during a potential power outage.
Things to Consider in Dell PowerEdge RAID Configuration
For a reliable and sustainable storage infrastructure, be sure to follow these rules:
- Use the Same Disk Types: Never mix SAS, SATA, and NVMe disks within a RAID group. Speed and capacity differences lead to instabilities. To learn about the advantages of NVMe disks, you can review our Dell PowerEdge NVMe Disk Installation and Advantages article.
- Define a Hot Spare Disk: Configure at least one "Global Hot Spare" disk so that the system automatically starts the rebuild process without waiting during a disk failure.
- Do Not Neglect Firmware Updates: Regularly update the firmware versions of the PERC controller and disks. Many disk failures and performance losses stem from outdated firmware versions.
Professional Installation and Support
Proper planning of server hardware and designing the RAID architecture to suit the workload requires expertise. Incorrect configurations can cause data loss and performance bottlenecks in the future.
With the professional services we offer within the scope of Hardware and Software Solutions, we plan the server infrastructure your business needs from end to end. From the mounting of your Dell PowerEdge servers to the operating system installation, we carry out all processes in accordance with best practice standards with our Server Installation, Configuration and Commissioning service.
Frequently Asked Questions
What happens if a disk fails in a RAID-configured system?
In a redundant RAID structure (RAID 1, 5, 6, 10), the system continues to run when a disk fails. When you replace the faulty disk while the system is running (hot-plug), the PERC automatically starts the rebuild process.
Is RAID configuration different for SSD disks?
Yes. Since SSD disks have very high write speeds, they can hit processor limits in structures that require parity calculations like RAID 5/6. RAID 10 or RAID 1 is usually preferred for SSDs.
Can the RAID level be changed while the server is running?
Dell PERC controllers support "Online Capacity Expansion" and "RAID Level Migration" features. However, these operations are risky, and a full backup must be taken before the operation.
Conclusion
Choosing the right RAID level on your Dell PowerEdge servers and optimizing PERC cache settings according to your workload maximizes the number of operations per second (IOPS) while minimizing your risk of data loss. Always take care to balance redundancy and performance when designing your infrastructure.
To receive professional support in your servers' RAID configuration, performance optimization, and hardware procurement processes, you can contact us.