HyperScale provides persistent storage for Linux-based container ecosystems. Storage volumes are not tied to the lifecycle of the container that they are attached to and persist across container restarts or terminations and are therefore ideal for long term data management. HyperScale uses a multi-tier storage architecture. The primary storage tier comprises of computes nodes that provide high performance storage for the containers. A secondary storage tier comprises of separate data nodes that manage off-host data services such as snapshots and backups. The application I/Os are primarily served from a high performing storage tier used as a cache that is periodically flushed to a lower performing storage tier. Real time replication of container writes to peer nodes coupled with the ability to take snapshots ensures lossless recovery of application data in the event of a disaster.
High availability of data
HyperScale preserves the state of the containers using persistent storage volumes. With persistent volumes, there is no loss of data when containers fail; data is always available no matter where the containers are scheduled in the container ecosystem. High availability of data is one of the prime considerations when planning for business continuity in case of disaster. For example, certain workloads, especially containerized enterprise database workloads require high resiliency of data. Without persistent storage, applications cannot reliably maintain a consistent state across ephemeral containers instances.
Predictable performance
With HyperScale, you can define storage Quality of Service (QoS) policies to ensure consistent storage performance specially in situations where there is contention for resources. You can configure the minimum and maximum storage performance goals in terms of IOPS. The QoS policies are honored by controlling the workload access to storage resources and by ensuring that a single container does not consume all the storage impacting the QoS policies of other containers in the ecosystem. The deployment of multiple workloads in the same cluster contending for the same resources can lead to "noisy neighbor" effects, adversely impacting application performance.
HyperScale ensures that fair storage bandwidth is provided to all the containers in the ecosystem and QoS policies are honored even in over-provisioned ecosystems. If policies cannot be met, alerts are generated.
Data protection
HyperScale isolates the data from a container to retain the benefits of adopting containerization. Data management is distinctly separate from the container lifecycle. With its multi-tier storage architecture, HyperScale provides real time fault recovery and continuous data protection. High-performance I/Os are isolated from other data services such as data replication, data backups, and volume snapshots.
You can configure the data resiliency factor and the data protection level when you configure the storage policies. The resiliency factor defines the availability of data. For example, a resilience factor of two indicates that the data is replicated to two additional nodes. To protect data, you can define the number of snapshots that are retained at any given point in time for the volume.