PowerScale: Cluster Capacity Usage, Raw compare to Available
Summary: The goal of this article is to provide some guidance about when to use Raw capacity and when to use Available capacity.
Instructions
There are many guides and documentation that mention cluster disk space, but they do not state whether to use Raw capacity or Available capacity. For example:
- The Upgrade Planning and Process Guide states that cluster capacity and
/ifsshould be at or below 90% (technical specifications). - Cluster Capacity Management documentation states that cluster capacity should stay below 95%~98%, or issues can be seen (Best Practices).
Raw and Available capacities are seen in the using the below command output. Depending on the cluster configuration, the output may be on a per node pool or per cluster basis.
isi status -v
Raw disk space is the total capacity of the cluster or node pool, including the Virtual Hot Spare (VHS)* in default settings.
*VHS is configurable and can vary in size depending on cluster hardware configuration and administrative preferences. Typically, it is configured as either a percentage of total storage (0-20%) or as a number of virtual drives (1-4). By default, it is set to one drive. Review the OneFS Users Guide for more information about configuring the VHS.
Available capacity is the free, unused disk space capacity that can be written to. This does not include the disk space that VHS consumes. For example:
Node Pool Name: a300_300tb_1.6tb-ssd_96gb Pool Storage: HDD SSD Storage Size: 2.1P (2.1P Raw) 0 (0 Raw) VHS Size: 38.1T Used: 1.1P (52%) 0 (n/a) Avail: 1012.7T (48%) 0 (n/a)
Raw Disk space capacity is used:
- For upgrades
- For
FlexprotectorFlexprotectlin - For spillover of the node pool (OneFS 9.5 provides a configurable limit)
Available Disk space capacity is used for:
- Available disk blocks to write to
AutobalanceandAutobalancelinjobs to balance/ifsdata per node pool**- OneFS operational writes and storage of configuration files
- Client, User, and Application data (That is File Storage.)
/root /var /var/crashpartitions are not used in the VHS allocation.
In most situations, whenever documentation says disk space, it implies Available Disk space.
** Autobalance or Autobalancelin jobs typically balance nodes to within 5% of each other.