ONTAP 9

to Japanese version

Cluster storage

The current iteration of ONTAP was originally developed for Fujitsu’s scale out cluster storage architecture. This is the architecture you typically find in datacenter implementations of ONTAP. Because this implementation exercises most of ONTAP’s capabilities, it’s a good place to start in understanding the concepts that inform ONTAP technology.

Datacenter architectures usually deploy dedicated ETERNUS HX or ETERNUS AX controllers running ONTAP data management software. Each controller, its storage, its network connectivity, and the instance of ONTAP running on the controller is called a node.

Nodes are paired for high availability (HA). Together these pairs (up to 12 nodes for SAN, up to 24 nodes for NAS) comprise the cluster. Nodes communicate with each other over a private, dedicated cluster interconnect.

Depending on the controller model, node storage consists of flash disks, capacity drives, or both. Network ports on the controller provide access to data. Physical storage and network connectivity resources are virtualized, visible to cluster administrators only, not to NAS clients or SAN hosts.

Nodes in an HA pair must use the same storage system model. Otherwise you can use any supported combination of controllers. You can scale out for capacity by adding nodes with like storage system models, or for performance by adding nodes with higher-end storage system.

Of course you can scale up in all the traditional ways as well, upgrading disks or controllers as needed. ONTAP’s virtualized storage infrastructure makes it easy to move data nondisruptively, meaning that you can scale vertically or horizontally without downtime.

scale out for capacity and performance
Top of Page