You need to plan the configuration of the file controller according to actual business requirements, and determine whether to enable the maximum bandwidth limit.
Note:
The configuration of the file controller is related to the amount of resources that the nodes in the cluster can provide. According to the principle of high availability, the file controller will be distributed on different nodes of the ACOS (AVE) cluster. Therefore, before planning the file controller, please understand the number of nodes in the ACOS (AVE) cluster, the number of vCPUs that can be exclusively used on each node, and the memory capacity of each node.
Each file controller will occupy 240 GiB of storage space. Please ensure that there is enough available storage space in the ACOS (AVE) cluster to which it belongs.
Quantity:
Each file storage cluster requires at least 3 file controllers and allows up to 16 file controllers. The number of file controllers should not exceed the number of nodes in the ACOS (AVE) cluster that can provide sufficient vCPUs and memory for a single file controller. Therefore, the limitation on the number of file controllers is as follows:
When the number of qualifying nodes in the ACOS (AVE) cluster is ≤ 16, 3 ≤ number of file controllers ≤ number of qualifying nodes in the ACOS (AVE) cluster.
When the number of qualifying nodes in the ACOS (AVE) cluster is > 16, 3 ≤ number of file controllers ≤ 16.
More file controllers can provide greater bandwidth and handle more concurrent access. The number of file controllers can be increased as needed after deployment. Please plan the number of file controllers according to actual business requirements.
vCPU:
Each file controller has CPU exclusive enabled by default, requiring at least 4 exclusive vCPUs and allowing up to 240 vCPUs.
Memory:
Each file controller requires at least 8 GiB of memory.
The file storage network and file access network may have high throughput. When these two networks share a VDS with other networks, please set the max bandwidth for the file controller to prevent the VM networks corresponding to these two networks from occupying bandwidth of other VM networks on the same VDS.
