Storage pool: Logical storage resource pools can be tagged into performance, local and capacity types.

Storage Policies: The storage strategy refers to the rules and methods that administrators use to create and manage the storage of data on virtual storage within the Sangfor platform. It involves factors such as the number of data copies, stripe count, etc., to ensure the security and availability of the data.

VM disk: Logical disk used by the virtual machine.

Disk group: A disk group refers to a structure formed by pairing cache disks and data disks in a specific relationship under a hybrid hard disk configuration. Each disk group consists of 1 SSD (Solid State Drive) and n HDDs (Hard Disk Drives), where the maximum value of n is 7. The SSD in the disk group is primarily used to provide caching acceleration for the underlying HDDs.

Cache Disk: The cache disk is used to improve the I/O performance of virtual storage. Solid-state drives (SSDs) are typically used as cache disks, leveraging the high-speed read/write advantages of SSDs to enhance the overall I/O performance of the storage.

Data Disk: The data disk serves as the hardware carrier for the underlying data of virtual storage. It is generally made of high-capacity, highly reliable enterprise-grade hard disk drives (HDDs).

Stripe: Striping is a technique used in the traditional RAID 0 mode. The general definition of striping technology is to split a continuous block of data into many small data blocks and then store them concurrently across different physical hard drives. This allows for maximum I/O concurrency when writing or reading data, thereby achieving excellent performance.

Stripe Depth (also known as Stripe Size): The size of the incremental data blocks in the aSAN striping process. The default stripe size for aSAN is set to 128K.

Number of Stripes: The default number of stripes in aSAN adapts based on the current storage volume structure. To ensure optimal I/O performance, the default value for the number of stripes will not exceed the number of data disks on any physical host in the storage volume. Typically, the default number of stripes is set to 6. If the number of data disks on any physical host in the storage volume is less than 6, the number of stripes will be set to the minimum number of data disks available on that host. If the number of data disks on the physical host exceeds 6, the number of stripes can be customized but must not exceed the number of data disks on any physical host in the storage volume. It is generally recommended not to modify the default value.

LUN: A LUN represents a logical volume of storage space on the target.

Replica: Replica is a redundancy technology that improves data reliability by storing multiple copies of the data. The virtual storage replication module is responsible for ensuring the consistency of replicas and the synchronization between them. From the perspective of upper-layer services, the replicas managed by the underlying system are transparent, meaning the upper layers are unaware of the existence of replicas. Types of replicas include:

1. Two Replicas: The same data is saved in two copies. When multiple hosts are present, this can handle the failure of one host. The available storage space is approximately 1/2.
2. Three Replicas: The same data is saved in three copies. When the cluster has 5 or more hosts, this can handle the failure of two hosts. The available storage space is approximately 1/3, with the highest level of redundancy.
3. Arbitration Replica: In a cluster of 3 or more hosts or two nodes cluster with witness node, in addition to data replicas, there is also an arbitration replica. The arbitration replica is a special type of replica, primarily used to prevent data split-brain scenarios, and it occupies very little space.

1.  The locations where replicas are stored must adhere to the host mutual exclusion principle, meaning no two replicas can exist on the same host.
2.  Available storage pool space = Total mechanical disk space in the cluster / Number of replicas. Therefore, replicas reduce the actual available capacity.
3.  Under normal conditions, without faults or anomalies, the data of file replicas is always consistent, and there is no distinction between a "primary" and a "secondary" replica.
4.  The arbitration replica also requires the host mutual exclusion principle, so a storage volume with two hosts does not have an arbitration replica. Arbitration is supported only in clusters with 3 or more hosts or two nodes cluster with witness node.