{{ $t('productDocDetail.guideClickSwitch') }}
{{ $t('productDocDetail.know') }}
{{ $t('productDocDetail.dontRemind') }}
6.11.3
{{sendMatomoQuery("Sangfor Cloud Platform (SCP)","Virtual Storage Pool Creation")}}

Virtual Storage Pool Creation

{{ $t('productDocDetail.updateTime') }}: 2026-04-27

Because HCI is divided into ordinary clusters and stretch clusters, the process of creating virtual storage will differ between them. There are certain conditions that must be met before creating virtual storage. Please ensure the following conditions are satisfied (HCI 5.8.6 and later version).

  1. For large cluster projects, it is recommended to use multiple volumes. It is recommended that the number of nodes per volume be less than 12 and not more than 16 as far as possible. For example, 48 nodes in the cluster are recommended to be divided into four volumes and 12 nodes per volume.
  2. RAID should be configured as JBOD passthrough. If JBOD is not supported, disable RAID and use non-RAID.
  3. Hybrid Datastore three Hosts and above.

Condition 1: The sum of the data disk capacity between hosts, the highest capacity, cannot exceed 80% of the sum of all the remaining hosts. Example: host A 5TB, host B 5TB, host C 5TB, host D is not allowed to exceed 12TB.

Condition 2: The ratio of the cache disk capacity to the data disk capacity of each host should not be less than 5%. For example, host A configuration with two 480GB SSDs and ten 4TB HDDs is not allowed.

Condition 3: The ratio of each host’s cache disks and data disks should be within 1:6. For example, host A has two cache disks and fifteen data disks, which is not allowed.

Condition 4: It is allowed to have a data disk with different capacities in a single host, but the capacity gap of the largest data disk should not exceed two times the smallest data disk. For example, a 1TB disk and a 4TB disk can form virtual storage but do not form a 1TB disk and a 4TB disk in the same disk group. The capacity of a 4TB disk is four times larger than a 1TB disk, but the performance is the same as a 1TB disk, so a 4TB disk can easily become a performance bottleneck. The bigger the gap, the bigger the bottleneck.

文本

AI 生成的内容可能不正确。

The HDD in the host allows SAS disks and SATA disks to be mixed, but it will cause the performance of SAS disks to be equal to SATA disks (barrel effect).

The host is allowed to have a varying number of disks between disk groups. The main impact is that the write performance will drop slightly when the storage pool is almost full.

 

 

  1. All-Flash Datastore three Hosts and above

Condition 1: The sum of the data disk capacity between hosts, the highest capacity, cannot exceed 80% of the sum of all the remaining hosts. Example: host A 5TB, host B 5TB, host C 5TB, host D is not allowed to exceed 12TB.

Condition 2: If the cluster physical nodes is less < 5 nodes each node required 2 SSD disk, and each nodes need create 2 disk group.

  1. Two hosts scenario

Condition 1: The total capacity of the two hosts must be equal.

Condition 2: The ratio of the cache disk capacity to the data disk capacity of each host should not be less than 5%. For example, host A configuration with two 480GB SSDs and ten 4TB HDDs is not allowed.

Condition 3: The ratio of the number of cache disks to data disks of each host should exceed or equal to 1:6. For example, host A has two cache disks, and fifteen data disks are not allowed.

Condition 4: It is allowed to have a data disk with different capacities in a single host, but the largest data disk capacity is not allowed to exceed 50% of the capacity of the smallest data disk. For example, a 1TB disk and a 4TB disk can form a virtual storage datastore but do not form a 1TB disk and a 4TB disk in the same disk group. The capacity of a 4TB disk is four times larger than a 1TB disk, but the performance is the same as a 1TB disk, so a 4TB disk can easily become a performance bottleneck. The bigger the gap, the bigger the bottleneck.

文本

AI 生成的内容可能不正确。

When expanding the capacity of two hosts in a datastore by add host, the entire HCI cluster needs to be restarted expanding by disks do not need restart.