RAID 2 +

RAID (Redundant Array of Independent Disks) is a redundant array of independent disks. 

It
is proposed by the University of California, Berkeley in 1987. The
basic idea is to combine multiple independent physical hard disks into a
virtual logical hard disk by using related algorithms, thereby
providing a larger, higher, or higher data error tolerance function.

Background
of block virtualization and reconstruction of traditional RAID: The
overall performance of the application system deteriorates, and the long
reconstruction time and data loss risks increase. Traditional RAID is
limited by the number of hard disks. When the data capacity increases
sharply, enterprises cannot flexibly and flexibly allocate resources.
Therefore, the underlying data distribution architecture must be highly
flexible and scalable, with the increase of hard disk capacity, data
management based on hard disks is becoming more and more difficult.

Huawei
provides RAID 2.0+ technical support by designing and developing
traditional RAID, LUN virtualization, and block virtualization.

The following figure shows the principle of Huawei RAID 2.0+：

The following figure shows the logical objects of Huawei RAID2.0+ software：

1. Disk
Domain is a combination of disks (which can be all disks in the
system). After the disks are combined and reserved for hot spare
capacity, the provides storage resources for the storage pool.

1) and HVS series storage systems can have one or more disk domains.

2). Multiple storage pools (Storage Pool) can be created in a disk domain.

3). Disks in a disk domain can be SSDs, SAS disks, or NL-SAS disks.

4). Different disk domains are isolated from each other, including fault domains, performance, and storage resources.

  2. Storage pool is a storage container. 

The
storage space used by all application servers comes from the storage
pool. A storage pool is created based on a specified disk domain. The
Chunk (CK) resources can be dynamically allocated from the disk domain,
and the Chunk Group (CKG) is composed of each storage level (Tier). RAID
protected storage resources.

Tier
is a storage tier, a collection of similar storage media in a storage
pool, used to manage storage media of different performances to provide
different storage space for applications with different performance
requirements. The storage pool can be divided into multiple Tiers
depending on the type of hard disk.

1).
Creating a storage pool specifies the storage tier (Tier) type that the
storage pool divides from the disk domain and the RAID policy and
capacity for that type.

2). HVS series storage systems support RAID5, RAID6 and RAID10.

3). The capacity layer is composed of a large-capacity NL-SAS disk. The RAID policy recommends using a double-check mode RAID6.

  3. Disk
Group (DG) is a set of disks of the same type in a disk domain. The
disk type can be SSD, SAS, or NL-SAS. The HVS series automatically
divides one or more Disk Group (DG) in each disk domain based on the
number of disks of each type.

1). One Disk Group (DG) contains only one type of hard disk.

2). Multiple CKs of any CKG come from different hard disks of the same Disk Group (DG).

  4. Logical Drive (LD) is a logical disk that is managed by the HVS series storage system. It corresponds to a physical disk.

 
   5.    CK is short for CK. It divides the hard disk space in a
storage pool into several fixed physical spaces. The size of each
physical space is 64MB, which is the basic unit of the RAID.

 
     6.   The Chunk Group (CKG) is a logical storage unit that consists
of CKs from different disks in the same DG based on the RAID algorithm.
It is the minimum unit for allocating resources from a disk domain to a
storage pool.

1). CKs in a CKG come from disks in the same DG.

2). CKG have RAID properties (the RAID properties are configured on the tier).

3). CK and CKG are internal objects of the system and are automatically configured by the HVS storage system.

 
    7.    Extents are logical storage spaces of fixed sizes based on
CKGs. The size of extents is adjustable. It is the minimum unit (data
migration granularity) for hotspot data statistics and migration, and is
the minimum unit for applying for space and releasing space in a
storage pool.

1). An extent belongs to one volume or LUN.

2). The size of and extent can be set when a storage pool is created. The size cannot be changed after being created.

3). The extents of different storage pools can be different, but the extents of the same storage pool are of the same size.

 
     8.    In thin LUN mode, extents are further divided into
fine-grained blocks based on the fixed size. These blocks are called
Grain. Thin LUNs are allocated based on Grain. LBAs in Grain are
consecutive.

             Thin LUNs are mapped to LUNs in the unit of Grain. For Thick LUN, this object does not exist.

 
     9.    A volume is an internal management object of the system. A
volume object is used to organize all extents and Grain logical storage
units of the same LUN. It can dynamically apply for and release extents
to increase or decrease the actual space occupied by the volume.

A LUN is a storage unit that can be directly mapped to a host for reading and writing. A LUN is a volume object.

Huawei RAID2.0+ Technical Features:

Automatic
load balancing reduces the overall failure rate. Quick thin
reconstruction, improving the failure rate of dual disks; Quick thin
reconstruction, improving the failure rate of dual disks; Fault
self-detection and self-healing ensure system reliability.

Huawei RAID2.0+ Reliability:

Load sharing: RAID2.0+ enables hard disks to work more evenly, preventing hard disks from being overworked.

Robust
reconstruction: With the RAID2.0+ technology, more disks are used to
share the reconstruction load during reconstruction, reducing the
reconstruction workload of each hard disk and reducing the risk of hard
disk faults during reconstruction.

Fast
reconstruction: The RAID2.0+ greatly reduces the time window for
reconstruction. In this way, the system can be restored to the fault
tolerance state as soon as possible, thereby improving system
reliability.

Simplified
reconstruction: RAID2.0+ can detect the used space in the allocated
space through metadata. Therefore, only used space is reconstructed
during reconstruction, which reduces the amount of data to be
reconstructed, shortens the reconstruction time, and reduces
reconstruction risks.

Self-detection and
self-healing: The RAID2.0+ uses the distributed hot spare space. When
the system detects a fault, the system automatically starts
reconstruction as long as the disk has free space (CK). This improves
reliability and reduces management costs.

Invalid
data volume: If a traditional RAID group fails, all data in the RAID
group is affected. When multiple disks fail on the RAID2.0+, only the
data associated with the failed disks becomes invalid. Most data can
still be accessed. The amount of invalid data is reduced by order of
magnitude compared with the traditional RAID.

Huawei RAID2.0+ dual-disk failure analysis:

RAID technology is the basis of storage data protection. It is still the fault tolerance capability of RAID.

For
RAID 5, the number of faults that can be tolerated is 1. For
traditional RAID, the unit is hard disk. For RAID2.0+, the unit is
blocked.

For RAID 6, the number of faults that can be tolerated is 2.

Therefore,
if a dual-parity protection type, such as RAID 6, is used, both
traditional RAID and block-based RAID2.0+ will not be lost when a
dual-disk failure occurs.

If
RAID5 is used, for traditional RAID, double disk failure will result in
data loss, while HVS series storage system adopting RAID2.0+ technology
will not have two failed blocks in each CKG as long as the dual disk
fails. (CK), then the data will not be lost.