Raw capacity is the total capacity on all the SSDs assigned to a Weka system cluster, e.g., 10 SSDs of 1 terabyte each have a total raw capacity of 10 terabytes. This is the total capacity available for the Weka system. This will change automatically if more hosts or SSDs are added to the system.
Net capacity is the amount of space available for user data on the SSDs in a configured Weka system. It is based on the raw capacity minus the Weka filesystem overheads for redundancy protection and other needs. This will change automatically if more hosts or SSDs are added to the system.
The stripe width is the number of blocks that share a common protection set, which can range from 3 to 16. The Weka system has a distributed, any-to-any protection. Consequently, in a system with a stripe width of 8, many groups of 8 data units spread on various hosts protect each other (rather than a group of 8 hosts forming a protection group). The stripe width is set during the cluster formation and cannot be changed. Stripe width choice impacts performance and space.
The protection level is the number of additional protection blocks added to each stripe, which can be either 2 or 4. A system with a protection level of 2 can survive 2 concurrent failures, while system data with a protection level of N+4 is protected against any concurrent 4 host/disk failures, and its availability is protected against any 4 concurrent disk failures or 2 concurrent host failures. A large protection level has space and performance implications. The protection level is set during the cluster formation and cannot be changed.
A failure domain is a group of Weka hosts, all of which can fail concurrently due to a single root cause, such as a power circuit or network switch failure. A cluster can be configured with explicit or implicit failure domains. For a system with explicit failure domains, each group of blocks that protect each other are spread on different failure domains. For a system with implicit failure domains, the group of blocks is spread on different hosts and each host is a failure domain. Additional failure domains can be added, and new hosts can be added to any existing or new failure domain.
Hot spare is the number of failure domains that the system can lose, undergo a complete rebuild of data, and still maintain the same net capacity. All failure domains are always participating in storing the data, and the hot spare capacity is evenly spread within all failure domains.
The higher the hot spare count, the more hardware required to obtain the same net capacity. On the other hand, the higher the hot spare count, the more relaxed the IT maintenance schedule for replacements. The hot spare is defined during cluster formation and can be reconfigured at any time.
After deducting the capacity for the protection and hot spares, only 90% of the remaining capacity can be used as net user capacity, with the other 10% of capacity reserved for the Weka filesystems. This is a fixed formula that cannot be configured.
Provisioned capacity is the total capacity assigned to filesystems. This includes both SSD and object store capacity.
Available capacity is the total capacity which can be used for the allocation of new filesystems, which is net capacity minus provisioned capacity.
The net capacity of the Weka system is obtained after the following three deductions performed during configuration:
Level of protection required, i.e., the amount of storage capacity to be dedicated for system protection.
Hot spare(s), i.e., the amount of storage capacity to be set aside for redundancy and to allow for rebuilding following a component failure.
Weka filesystem overhead, in order to improve overall performance.