EverON High Availability Clustering

EverON High Availability Clustering

Continuous data availability and reliable performance are prerequisites for operating in today's business critical environments. EverON OS provides N-way high availability clustering capabilities to allow multiple ONStor NAS systems to be configured as an Active-Active cluster. EverON High Availability Clustering (EverON HA) allows administrators to scale performance and capacity independently while meeting stringent levels of uptime and availability.

• Active-Active, multi-node clustering
• Automated & transparent failover
• Stretch cluster for campus wide clustering
• Filer groups for selective failovers
• Scale capacity and performance independently

• Optimal hardware utilization with no requirements for standby hardware
• Provides continuous data availability to end users and applications
• Survive potential data center or building disasters in a campus
• Match availability to application needs
• Cost effective scaling while maintaining stringent levels of uptime

Continuous Data Availability

Under normal operation, each NAS system in the cluster provides data services to applications with EverON HA service monitoring the health of all NAS systems in the cluster. In the event of a failure of one or more ONStor NAS systems or if a NAS system is taken offline, EverON HA initiates a takeover operation of the data services of the failed NAS system(s). Takeover is automatic and transparent to end users and applications. The data services of the failed system are transferred non-disruptively to the remaining NAS systems in the cluster. During and after the takeover operation the data services on the other NAS systems in the cluster are never impacted.

Administrators take advantage of EverON HA capabilities to provide continuous data availability during periods of upgrades and maintenance. To upgrade one or more systems in a cluster, an administrator uses a rolling methodology wherein data services on each NAS system in the cluster are failed over with EverON HA at the beginning of an upgrade and are failed back after a successful upgrade.

EverON HA clustering offers customers the flexibility and choice of matching data services with requisite availability and uptime needs. Multiple ONStor NAS systems can be grouped together in filer groups to facilitate failover of data services between the systems in the group. In addition, EverON HA enables administrators to stretch a cluster by placing the NAS systems in different physical locations (data centers) in a campus to survive potential disasters to a data center or a building. EverON HA uses standard IP networking for the cluster interconnect network making it inexpensive to deploy campus wide clustering.

Scaling performance and capacity independently

ONStor Clustered NAS systems provide transactional guarantee of data persistence when applications write data to the system. EverON OS uses a complete write-through design for storage i/o. Data written to an ONStor NAS system are journaled on a stable storage pool before acknowledgement of data write is sent back to applications. This avoids the complexity of battery-backed NVRAM and having to mirror NVRAM between NAS systems in a cluster. Since transactional data is written to a stable storage pool and the pool is shared across all NAS systems in the cluster, the EverON HA cluster can scale to a large number of NAS systems. Today, customers can scale an EverON HA cluster to 8 ONStor NAS systems. The flexibility offered by EverON HA clustering enables customers to scale performance and capacity independently. To increase aggregate performance of the cluster, simply add new ONStor NAS systems to the cluster. Adding a NAS system to a cluster is non-disruptive and is automated by the EverON operating systems' non-stop cluster management framework.