What's new in ONTAP 9.19.1 (admin's guide)

Data protection

The replication story is the biggest chunk of 9.19.1, and it is mostly SnapMirror active sync growing into new workloads and topologies:

• SnapMirror active sync: transparent application failover for AIX. Symmetric active/active now supports transparent failover with zero RPO for AIX on 2-node clusters. The AIX host keeps both copies usable and does not wait for a manual or scripted failover step.
• SnapMirror active sync: NAS at the SVM level. NAS workloads are supported at the SVM level using NFS and SMB on AFF clusters (2-node only) or AFX clusters (4-node only), with read-write access only on the primary. The topology constraints are strict — check them before planning a NAS DR pair.
• SnapMirror cloud: 100 S3 buckets. The number of S3 buckets supported per SnapMirror cloud relationship is raised to 100, up from the previous limit.
• SnapMirror synchronous: tamperproof snapshot locking. You can apply tamperproof snapshot locking to SnapMirror synchronous — protection copies an attacker (or an admin with bad intent) cannot silently delete.
• SnapMirror synchronous: scheduled-snapshot replication. Scheduled snapshots can now be replicated to the destination volume, closing the gap where only explicitly created snapshots went over the wire.

Admin take: if you have been holding AIX or NAS workloads off active sync because only SAN on ASA qualified, 9.19.1 removes that reason — but the 2-node/4-node and primary-read-write limits mean the design review matters more, not less. The active sync reference page covers the architecture and caveats.

SAN

• Active-active multipathing on AFF systems. AFF now supports active-active SAN multipathing: all paths within the HA pair that owns the LUN or namespace are active-optimized, so on storage failover the host does not need to wait for the ALUA or ANA transition of failover paths to resume I/O. NetApp's framing (confirmed in the NetApp community announcement of September 2026): this extends symmetric active-active SAN pathing from ASA into unified AFF starting with 9.19.1, giving SAN customers the same non-disruptive path resiliency and load balancing on AFF that was previously available only on ASA — aimed at consolidating mission-critical SAN, NAS, and mixed workloads on one platform.
• Direct-attach FC without switches. FC switches are no longer required to connect FC and FC-NVMe hosts: hosts can connect directly to FC adapter ports on AFF, ASA, and FAS systems. Especially useful at remote sites and in simple hybrid environments.
• Automatic SAN LIF failover extended. Automatic LIF failover is now supported for iSCSI LIFs in active-active and active-active-scsi-only multipath configurations and for NVMe/TCP LIFs in active-active configurations, on AFF and ASA systems.
• Foreign LUN Import (FLI) for iSCSI. FLI now supports iSCSI as the backend connectivity protocol, widening the array-migration on-ramp.

Admin take: the AFF active-active change is the one to re-validate: anyone running SAN workloads on AFF with ALUA/ANA-based expectations should re-test multipath and failover behavior after upgrading — the pathing model itself has changed.

Networking

• NFSv4.2 managed per-SVM. NFSv4.2 support is now managed independently of NFSv4.1, so each SVM can advertise or suppress NFSv4.2 as needed. By default NFSv4.2 is enabled when a new NFS service is created, unless an administrator explicitly disables it.
• NAS LIF failover policy broadened. The ONTAP system-defined failover group policy now dynamically expands the failover targets for NAS data LIFs to potentially include any available port in the broadcast domain on any available node in the cluster. Before 9.19.1, LIFs with this policy could fail over to only one other node — and if that node was out of quorum, the LIF would not fail over even when other nodes were available.
• TCP loss recovery: PRR and RACK. ONTAP integrates PRR (proportional rate reduction) and RACK (recent acknowledgment) into its TCP stack for lossy, high-latency WAN links. PRR is suitable for all networks and is enabled by default; RACK suits congested links around 25 Gbps and is disabled by default (opt-in).

Admin take: the failover-policy change is a genuine behavior fix for the "out-of-quorum stuck node" failure mode — but it also changes where your NAS data LIFs can land after a failure, so re-check failover-group expectations per broadcast domain. The NFSv4.2 default applies only to newly created NFS services; existing SVMs keep their current version settings until you change them.

NAS performance

• Small-directory listing boost. ONTAP 9.19.1 devotes more CPU to listing requests for small NAS directories (about 25,000 files, 2 MB or less). This helps when hundreds to thousands of clients list the same small directory at once, and reduces the need for workarounds such as local FlexCache volumes for high-client-count enumeration. The enhancement is automatic — no configuration required — and does not hurt performance at lower client counts.

Admin take: if you deployed FlexCache (or a load-spreading design) purely to survive thundering-herd directory listings, 9.19.1 is a reason to re-test whether that workaround is still needed.

S3 object storage

• Conditional writes and deletes with enforcement. You can attach conditions to object uploads (conditional writes) and object deletes (conditional deletes) to prevent overwrites and unintended deletions of objects. See the S3 on ONTAP guide for the object-store server context.
• Dual S3 user access keys. Cluster administrators can generate a second access key for each S3 user, so access survives rotation of the original key without downtime.

Admin take: conditional-write enforcement is the guardrail for workloads where a misbehaving client (or a bot) must never clobber objects — pair it with bucket policies rather than trusting client behavior.

Security & ARP/AI

• ARP/AI covers replicated volumes. Autonomous Ransomware Protection with AI now protects primary volumes in SnapMirror synchronous relationships (NAS and SAN) on FAS, AFF, and AFX systems, and in SnapMirror active sync SAN relationships on AFF and ASA r2 systems. Detection runs on the active read-write primary, where ARP/AI snapshots are created — ARP/AI snapshots are not replicated to the secondary.
• ARP/AI default restored on upgrade. Clusters upgraded to ONTAP 9.19.1 that previously had default ARP/AI enablement suppressed because of synchronous replication relationships will have it automatically restored. A pre-update warning tells you this is coming and gives you the option to disable it.
• NFS over TLS 1.3 + mutual TLS. NFS over TLS uses TLS 1.3 to encrypt NFS traffic in transit on NFS-enabled SVMs, with optional per-LIF mutual TLS host authentication and export-policy controls that can deny non-TLS TCP access for matching clients.
• TLS offload to NICs. You can configure TLS offload so supported Ethernet cards handle TLS encryption and decryption, cutting CPU overhead and improving post-handshake performance.
• ARP "surge" presentation removed from System Manager. System Manager no longer shows ARP workload characteristics (baseline and surge statistics) because surge activity was frequently misinterpreted as an actual ransomware attack. Detection behavior is unchanged; surge statistics remain available via the ONTAP CLI and REST API for advanced analysis.
• WebAuthn relying-party lockdown. A new -rp-domains parameter in the WebAuthn global settings command restricts the set of permitted relying-party domains accepted during WebAuthn (FIDO2) multi-factor authentication in System Manager.

Admin take: two follow-ups after upgrading: re-baseline ARP alerting on CLI/API output if your team watched the System Manager surge chart, and expect (or deliberately decline) the ARP/AI auto-restore warning during pre-update checks on clusters with synchronous replication.

Storage management (SRM & System Manager)

• Enhanced volume space reporting. Beginning with 9.19.1, volume show-space output includes the directory used for the space reporting. Scripts that parse this command's output should tolerate the new field.
• QoS throughput-ceiling settings now modifiable. You can now modify the default settings that control how QoS policy groups automatically raise the QoS throughput ceiling in response to short-term performance needs. These settings could not be modified in ONTAP 9.18.1 and earlier.
• Dedicated SVM dashboard in System Manager. A new dashboard lets SVM administrators monitor and manage their own SVMs — useful in tenanted ONTAP environments where the cluster administrator can hand each tenant an SVM dashboard.

Admin take: the QoS change is the sleeper item for shared clusters: automatic ceiling-increase behavior can now be tuned, which matters when noisy neighbors trigger ceiling raises that affect other workloads.

AFX & ASA r2

• AFX 2K requires 9.19.1. Per this site's release guide and hardware-compatibility notes, 9.19.1 is the minimum ONTAP train listed for AFX 2K systems — the platform floor, not an option.
• AFX has its own 9.19.1 release-notes cut (the AFX-1.5 update). Repo-sourced highlights: SnapMirror active sync for NAS on AFX clusters (SVM level, NFS/SMB, 4-node only, read-write on the primary), the SnapMirror cloud 100-bucket limit, and the PRR+RACK TCP work all apply; plus AFX-specific efficiency features — Cross File Sequential Read (CFSR) prefetching for video/M&E sequential-read workloads, Global Deduplication extending dedup scope from per-volume to the whole Storage Availability Zone (SAZ, on by default), and Dynamic Storage Efficiency auto-suspending inline/background dedup, compression, and compaction during high-volume write bursts to protect throughput.
• ASA r2 has its own what's-new page in NetApp's documentation — check it alongside the main notes if you run ASA r2.

Admin take: on AFX, dedup savings assumptions change: efficiency now computes at SAZ scope, and the auto-suspend behavior means dedup ratios dip during write bursts by design — not a fault.

What changed for existing clusters

The features above are opt-in or additive. These are the changes that alter behavior of clusters you already run the day after upgrade:

• ARP/AI enablement can come back on. Clusters that had default ARP/AI suppressed by synchronous replication relationships get the default restored on upgrade to 9.19.1 — the pre-update warning is your chance to decline.
• NAS LIF failover targets broaden. The system-defined failover group policy now allows failover to any available node/port in the broadcast domain, not one designated other node. Failure behavior improves; failover destinations become less predictable.
• PRR is on by default in the ONTAP TCP stack. RACK stays off unless you enable it.
• NFSv4.2 default-on applies to new NFS services only. Existing SVMs are untouched until an administrator changes their NFS version settings.
• ARP surge is gone from System Manager (CLI/API only now) — re-baseline any dashboard that watched it.
• volume show-space gains a field (the reporting directory) — parse tolerant.
• QoS automatic ceiling-increase settings become modifiable (they were read-only in 9.18.1 and earlier).
• On AFX: Global Deduplication runs by default at SAZ scope and Dynamic Storage Efficiency may suspend efficiency work during write bursts.

Upgrade notes

9.19.1 is a feature release, and the release guide's warning applies: feature releases establish capabilities, but production planning must also select a supported P-release and check known issues, upgrade paths, platform support, and host interoperability. NetApp notes that the ONTAP 9 Release Notes (known issues, limitations, and upgrade cautions) require signing in with a NetApp account.

• Check the pair, not the train. Whether your current release can jump straight to 9.19.1 (or needs an intermediate stop) is pair- and platform-specific — work the supported upgrade paths guide and confirm with Upgrade Advisor and the official upgrade documentation before staging anything.
• Pre-flight is unchanged. The standard checks from the release guide still apply before any hop:

version -v
system node show -fields model,serialnumber,health,uptime
cluster show
storage failover show
system health alert show
cluster image show
cluster image validate -version <target-version>

• Pre-update checks now separate errors from warnings in automated output, so issues that must be resolved before upgrading are easier to spot — including the ARP/AI auto-restore warning described above.
• Rolling-upgrade mechanics are unchanged: one node at a time with automatic HA takeover, image install, and giveback — follow the rolling upgrade runbook for the full procedure, verification, and rollback steps.

Sources and boundaries

This synthesis is built from NetApp's official What's new in ONTAP 9.19.1 page (documented July 28, 2026), the AFX-specific 9.19.1 release notes, this site's gathered documentation summaries and news coverage of the release, the NetApp community announcement on symmetric active-active SAN pathing for unified AFF, and this site's release guide. It is community guidance, not a NetApp support statement. Feature availability can vary by platform and protocol — for known issues and upgrade cautions, sign in to the ONTAP 9 Release Notes and check Hardware Universe and the Interoperability Matrix for your exact configuration.