ONTAP Select & Edge Deployment Guide
Plan software-defined ONTAP for ROBO and edge sites: ONTAP Select architecture, ESXi and KVM requirements, HA, licensing, performance limits, migration, and the legacy Data ONTAP Edge name.
Names and current status
ONTAP Select is current software-defined ONTAP: nodes run as VMs on supported VMware ESXi or KVM hosts and are managed through ONTAP Select Deploy. Current docs do not list direct AWS/Azure deployment; use Cloud Volumes ONTAP there.
Data ONTAP Edge was a legacy predecessor. There is no current separately documented free “ONTAP Edge” edition. Select offers a time-limited evaluation deployment; verify current licensing and capacity limits.
Single-node, HA and scale-out
| Model | Fit | Key point |
|---|---|---|
| Single node | Lab, dev/test, small branch | No storage-node HA |
| Two node | ROBO/edge HA | Deploy VM continuously provides HA quorum |
| 4/6/8 node | Larger SDS cluster | Separate hosts, consistent hypervisor release |
| MetroCluster SDS | Stretched two-node design | Strict support requirements |
Planning baseline
Current docs list Intel Xeon; AMD is unsupported. The small-node baseline is at least six physical cores with four reserved and 24 GB RAM with 16 GB reserved. A single node needs two 1GbE minimum (two 10GbE recommended); two-node needs four 1GbE or one 10GbE minimum (two 10GbE recommended). Medium/large profiles require more CPU, memory and license capability.
Hardware RAID, software RAID and supported external storage have different rules. Validate the exact Select, Deploy, hypervisor, server, NIC and storage combination in current docs and the Interoperability Matrix.
Deploy and ONTAP management
deploy> host register -name edge-esx01 -mgmt-server 192.0.2.21
deploy> host show -name edge-esx01
deploy> cluster create -help
deploy> cluster showcluster::> system node show
cluster::> network interface show
cluster::> storage aggregate show
cluster::> system license showDeploy owns VM lifecycle; System Manager and the ONTAP CLI manage data. Never modify a Select VM directly unless NetApp Support directs it.
Performance, uses and limitations
Select performance depends on reserved vCPU, host scheduling, NICs, RAID and storage latency. Unlike physical AFF/FAS/ASA, it uses general-purpose servers and a hypervisor, without the integrated controller/media path. Benchmark workload and failure states.
- Good fits: ROBO, tactical edge, lab, dev/test and qualified server reuse.
- Scale is bounded by licensed node profiles and documented limits.
- Software RAID configurations have special restrictions; VMware vMotion, HA and DRS are not supported there.
- Supportability depends on the complete validated stack.
Platform comparison
| Dimension | AFF/FAS/ASA | ONTAP Select | Data ONTAP Edge |
|---|---|---|---|
| Status | Current hardware | Current SDS product | Legacy predecessor |
| Runtime | NetApp controller | Supported ESXi/KVM VM | Historical virtual appliance |
| HA | Integrated HA | Single or supported multi-node | Archived docs only |
| License | Platform entitlement | Select license/evaluation | Historical |
Also read Select supportability and physical platforms.
Replication and migration
Use supported SnapMirror relationships between Select and physical ONTAP, subject to release compatibility, licensing and cutover planning.
snapmirror create -source-path edge_svm:vol1 -destination-path dc_svm:vol1_dr -type XDP
snapmirror initialize -destination-path dc_svm:vol1_dr
snapmirror update -destination-path dc_svm:vol1_dr
snapmirror break -destination-path dc_svm:vol1_dr