Hyperconverged Infrastructure Storage Domain Segmentation

The patent divides the cluster into multiple storage domains, where each domain is a self-contained unit with its own set of host computing devices, virtualized machines, and datastores. This segmentation allows the cluster to scale by adding entire storage domains rather than individual hosts, reducing the complexity increase associated with growth. Each storage domain operates semi-independently, preventing complexity from propagating across the entire system.

By segmenting the cluster into multiple storage domains with dedicated datastores for each domain, the patent eliminates the performance degradation caused by shared resource contention. Each storage domain has its own isolated datastore, ensuring that increased numbers of hosts do not lead to increased system latency or decreased bandwidth, as resources within each domain remain dedicated and uncontended.

The patent implements fault isolation by creating separate storage domains that act as independent fault boundaries. When a fault occurs in one storage domain, it remains contained within that domain and does not propagate to other domains. This segmentation enables effective fault isolation and recovery even as the cluster grows to include thousands of hosts, directly addressing the deterioration of reliability capabilities.

The patent segments the common datastore into multiple domain-specific datastores, with each datastore dedicated to a specific storage domain. This segmentation maintains high resource utilization within each domain while preventing the performance degradation that occurs in shared datastores as cluster size grows. Each datastore serves only the hosts within its domain, eliminating cross-domain resource contention.