Intel® Virtual RAID on CPU (Intel® VROC) RAID Levels Support List

Intel® Virtual RAID on CPU (Intel® VROC) is an enterprise RAID solution that unleashes the performance of NVMe* SSDs. Intel® VROC is enabled by a feature in Intel® Xeon® Scalable processors called Intel® Volume Management Device (Intel® VMD), an integrated controller inside the CPU PCIe root complex. NVMe* SSDs are directly connected to the CPU, allowing the full performance potential of fast storage devices to be realized. Intel® VROC enables these benefits without the complexity, cost, and power consumption of traditional hardware RAID host bus adapter (HBA) cards placed between the drives and the CPU.

The supported RAID levels by Intel® VROC are described below.

RAID 0 (Striping)

RAID 0 uses the read/write capabilities of two or more drives working in parallel to maximize the storage performance of a computer system. The following information provides an overview of the advantages, the level of fault tolerance provided, and the typical usage of RAID 0.

• Drives Supported: Two minimum.
• Advantage: High transfer rates.
• Fault Tolerance: None. If one drive fails all data will be lost.
• Application: Typically used in desktops and workstations for maximum performance for temporary data and high I/O rate. It also should be noted that although RAID 0 can be scaled to many drives there is a performance sweet spot specific for each implementation.

RAID 0 combines at least two (up to the max number) of drives supported by SATA(8)/sSATA(6)/tSATA(6) controller on the platform or up to 24 NVMe* drives so that all data is divided into manageable blocks called strips. The strips are distributed across the array members on which the RAID 0 volume resides. This improves read/write performance, especially for sequential access, by allowing adjacent data to be accessed from more than one drive simultaneously. However, data stored in a RAID 0 volume is not redundant. Therefore, if one drive fails, all data on the volume is lost.

The RAID 0 volume appears as a single physical drive with a capacity equal to the number of drives in the volume times the size of the smallest drive in the volume.

The maximum number of drives supported in a RAID 0 is up to the maximum number of drives supported by the platform.

RAID 1 (Mirroring)

RAID 1 volumes contain two drives where the data is copied to both drives in real time to provide data reliability in the case of a single disk failure. When one disk drive fails, all data is immediately available on the other drive without any impact to the integrity of the data. The following information provides an overview of the advantages, the level of fault tolerance provided, and the typical usage of RAID 1.

• Drives Supported: Two maximum.
• Advantage: Redundancy of data. One drive may fail, but data will continue to be accessible. A rebuild to a new drive is recommended to maintain data redundancy.
• Fault Tolerance: Excellent. Drive mirroring means that all data on one drive is duplicated on another drive.
• Application: Typically used for smaller systems where the capacity of one disk is sufficient and for any application(s) requiring very high availability.

RAID 1 combines two drives so that all data is copied concurrently across the array members that the RAID 1 volume resides on. In other words, the data is mirrored across the drives of the RAID 1 volume. This creates real-time redundancy of all data on the first drive, also called a mirror. RAID 1 is usually used in workstations and servers where data protection is important.

The RAID 1 volume appears as a single physical drive with a capacity equal to the smaller drive.

The maximum number of drives supported in a RAID 1 volume is two drives.

RAID 5 (Striping with Parity)

RAID 5 volumes contain three (minimum) or more drives where the data and parity are striped across all drives in the volume. Parity is a mathematical method for recreating data that was lost from a single drive, which increases fault tolerance. If there are N drives in the RAID 5 volume, the capacity for data would be N-1 drives. For example, if the RAID 5 volume has 5 drives, the data capacity for this RAID volume consists of 4 drives. The following information provides an overview of the advantages, the level of fault tolerance provided, and the typical usage of RAID 5.

• Drives Supported: Three minimum.
• Advantage: High percentage of usable capacity and high read performance as well as fault tolerance.
• Fault Tolerance: Excellent. Parity information allows data to be rebuilt after replacing a failed drive with a new drive.
• Application: Typically used as storage of large amounts of critical data. As with RAID 0, although RAID 5 can be scaled to many drives, there is a performance sweet spot specific to each implementation.

RAID 5 provides the capacity of N-1 times the smallest size of the drives, where N >= 3 and:

• Less or equal to the maximum number of drives supported by SATA/sSATA/tSATA controller, or;
• Less or equal to 48 NVMe* drives.

So that all data is divided into manageable blocks called stripes. RAID 5 also stores parity, a mathematical method for recreating lost data on a single drive, which increases fault tolerance. The data and parity are striped across the array members. The parity is striped in a rotating sequence across the members. Because of the parity striping, it is possible to rebuild the data after replacing a failed drive with a new drive. However, the extra work of calculating the missing data will degrade the write performance of the volumes. RAID 5 performs better for smaller I/O functions than larger sequential files.

RAID 5, when enabled with volume write-back cache combined with the built-in coalescer, will enhance write performance. This combines multiple write requests from the host into larger and more efficient requests, resulting in full stripe writes from the cache to the RAID 5 volume.

A three-drive RAID 5 will provide capacity twice the size of the smallest drive. The remaining space will be used for parity information.

The maximum number of drives supported in a RAID 5 is the maximum number of drives supported by the platform.

RAID 10 (Striping and Mirroring)

A RAID 10 volume uses four drives to create a combination of RAID levels 0 and 1. It is a striped set whose members are each a mirrored set. It provides a great balance between performance and excellent fault tolerance as it allows two drives to fail while still maintaining access to data, but has a low-cost effectiveness. The double degradation (two drives failure) support however, is limited to cases where drives from opposite mirrored set are failed. The following information provides an overview of the advantages, the level of fault tolerance provided, and the typical usage of RAID 10.

• Drives Supported: Four.
• Advantage: Combines the read performance of RAID 0 with the fault tolerance of RAID 1.
• Fault Tolerance: Excellent. Drive mirroring means that all data on one drive is duplicated on another drive.
• Application: Typically used for high-performance applications requiring data protection, such as video editing.

In a RAID 10 volume, the data is striped across a two-disk array forming a RAID 0 component. Each of the drives in the RAID 0 array is mirrored to form a RAID 1 component. This provides the performance benefits of RAID 0 and the redundancy of RAID 1.

The RAID 10 volume appears as a single physical drive with a capacity equal to the two smallest drives of the four-drive configuration (the only supported RAID 10 configuration). The space on the remaining two drives will be used for mirroring.

The maximum number of drives supported in a RAID 10 is four.