Unlike other NetApp ONTAP storage systems, the compute and storage elements of an AFX cluster are decoupled and joined through a switched network. Disk ownership is no longer tied to specific nodes which provides several benefits. For example, the compute and storage components of an AFX cluster can be expanded independently.

The physical aggregates are no longer available to the AFX storage administrator. Instead, AFX automatically manages the virtual capacity allocations for the nodes, as well as the RAID group configuration, when new storage shelves are added to the cluster. This design simplifies administration and provides an opportunity for nonspecialists to manage their data.

Because the storage nodes and shelves are decoupled with NetApp AFX, all storage capacity for the cluster is collected in a single pool known as a Storage Availability Zone (SAZ). The disks and shelves in a SAZ are available to all the storage nodes in an AFX cluster for read and write operations. In addition, all the cluster nodes can participate in disk rebuilds in the event of a failure. Refer to FAQ for AFX storage systems for more details.

NetApp AFX provides high and sustained bandwidth with ultra-low latency and so is designed for high performance NAS and object workloads. AFX uses the latest modern hardware as well as storage shelves capable of handling a high ratio of nodes to disks through its unique architecture. Scaling storage nodes beyond the typical 1:1 (node:shelf) ratio maximizes the possible performance profile of the disks to their edge limits. This design provides efficiency and storage density for your most critical applications.

Based on the decoupled storage nodes and shelves, an AFX cluster can be independently and nondisruptively expanded based on your application needs. You can add storage nodes to get more CPU and throughput or add shelves to get more storage capacity and disk performance. The NetApp AFX architecture brings new possibilities for the maximum size of your cluster. For the latest limits for the AFX cluster based on your ONTAP release, refer to the NetApp Hardware Universe.

NAS and object clients access volumes at the ONTAP cluster. You can relocate volumes across the nodes nondisruptively to achieve your capacity and performance balancing goals. With Unified ONTAP, a volume move is performed using SnapMirror technology which can take time and additional temporary capacity. But with AFX, a data copy operation is no longer needed within the shared Storage Availability Zone (SAZ). Instead, only the volume metadata is moved which dramatically improves performance. Refer to FAQ for AFX storage systems for more details.

NetApp AFX offers a number of enhancements for high availability (HA) configuration and processing. AFX removes the need to directly connect HA partner nodes and instead allows HA pairs to communicate over the internal cluster network. This design gives administrators the option of deploying HA pairs in separate racks or rows in a datacenter for added fault tolerance. In addition, the AFX zero copy mobility extends to HA failover scenarios. When a node fails, its volumes will failover to the HA partner to commit any remaining writes to disk. Then ONTAP balances the volumes evenly across all surviving nodes in the cluster. This means you no longer need to consider storage failover performance in the initial design of your data placement.