Upgrading to VCF 9.1 from vSphere + Aria + Dell Storage: The Storage-Specific Migration Guide

This is Part 12 of an architect’s read of VCF 9.1, and the third of four vendor-specific migration guides. Parts 10 and 11 covered NetApp and Pure. This one is for customers running Dell storage — PowerStore, PowerMax, or PowerFlex — with vSphere + Aria today.

Dell’s storage portfolio under VCF takes three architecturally distinct paths. PowerStore and PowerMax are traditional SAN arrays — customers using them with VCF land cleanly in the VMFS-on-FC + supplemental NVMe/TCP pattern. PowerFlex is software-defined and uses Dell’s SDC (Software-Defined Client) inside ESX hosts to consume storage — a different integration model that requires explicit permission from VMware and Dell. Then there’s the Dell Private Cloud offering, which packages VCF on Dell hardware as a managed solution and is gaining VCF 9.1 support in mid-2026.

Each path has its own migration mechanics. This piece is the architect’s walkthrough.

Source state: Dell storage + vSphere today

Three common source-state archetypes.

PowerStore-backed estate. PowerStore arrays providing block (FC, iSCSI, NVMe/TCP, NVMe/FC) and optional file (NAS) services. vSphere 7.x/8.x with vCenter. Dell Virtual Storage Integrator (VSI) or PowerStore vSphere plug-ins for in-vCenter operations. PowerProtect Data Manager for backup. RecoverPoint or PowerStore native replication for DR. Aria Operations with the Dell management pack.

PowerMax-backed estate. PowerMax arrays (typically all-flash NVMe) for mission-critical workloads. vSphere 7.x/8.x. Predominantly FC-attached. SRDF (Symmetrix Remote Data Facility) for synchronous or asynchronous replication. Some vVols deployment for fine-grained policy management. PowerProtect Data Manager and/or Data Domain for backup. Dell EMC Unisphere for storage management.

PowerFlex-backed estate. PowerFlex software-defined storage running on Dell servers, with SDC (Software-Defined Client) installed on ESX hosts. vSphere 7.x/8.x with VMFS-on-FC or NVMe/TCP as the access protocol. PowerFlex Manager for orchestration. Replication via PowerFlex native facilities. Often used for hyperscale or AI-class infrastructure.

What VCF 9.1 changes for Dell customers

VMFS on FC is supported as primary storage for the Management Domain. Like NetApp and Pure customers, Dell PowerStore and PowerMax customers benefit from VCF 9’s removal of the vSAN-only requirement for the Management Domain. A VCF 9.1 deployment can land natively on a PowerStore or PowerMax LUN via FC. No vSAN required. The implementation guides from Dell’s engineering blog cover both PowerMax and PowerStore on VCF 9 with FC for both Management and Workload Domains.

vVols termination date is now formal. Industry signals from the Dell side (Drew Tonnesen’s blog and Broadcom KB updates) point to VCF 9.3 as the formal termination release for vVols. Deprecation began in VCF 9.x, with 9.0 keeping support, 9.1 being a maintenance release with no vVols changes, and full removal coming in 9.3. PowerMax customers with vVols deployments need a migration plan within that timeline. Most Dell customers don’t use vVols heavily (it never had the depth of NetApp or Pure adoption on PowerMax / PowerStore), but where it exists, the migration must happen.

PowerFlex requires the SDC for vSphere 9. Dell PowerFlex 5.x completed certification for the SDC with vSphere 9, with SDC 5.1 supporting ESX 9.x. The certification of NVMe/TCP and the SDC for vSphere 9 means VCF 9 (which requires vSphere 9) is supported on PowerFlex. The SDC remains a custom install requiring partner-supported acceptance level on ESX hosts. PowerFlex SDC is changing how it integrates with ESXi at VMware’s request — the current implementation evolves; the architectural role of SDC continues.

NVMe/TCP is the recommended modern alternative. For PowerFlex customers, NVMe/TCP is an attractive alternative to the SDC-based path — standardised, well-supported, no special permissions. For PowerMax and PowerStore, NVMe/TCP also offers a modern protocol option alongside the dominant FC pattern.

VCF 9.1 is a normal upgrade for existing 9.0 Dell customers. Dell’s own commentary frames VCF 9.1 as a maintenance release for external-storage customers — the VCF Operations and VKS improvements matter, but the storage-specific impact is minimal versus 9.0. For PowerMax and PowerStore on 9.0, the move to 9.1 is straightforward. For PowerFlex on 9.0, the SDC may need a corresponding upgrade if a new ESX 9.1 build is published.

Dell Private Cloud with VCF 9.1 lands June 2026. Dell Private Cloud (DPC) deploying VMware with VCF 9.1 support is announced for June 2026. DPC can scale compute and storage independently — Dell claims up to 65% cost savings vs HCI. For customers considering a packaged Dell-managed VCF deployment, this is the timeline reference point.

Migration paths for Dell customers

Path A: PowerStore + Convergence on FC. PowerStore-backed estate using primarily VMFS on FC. vCenter and ESX upgrade to v9, convergence brings vCenter into VCF as the Management Domain on the existing PowerStore FC LUN. Datastores remain mounted. VSI plug-in continues to provide operational features. NVMe/TCP can be added as supplemental storage post-convergence.

Path B: PowerMax + Convergence with parallel vVols migration. PowerMax-backed estate. If no vVols are in use, Convergence path works cleanly. If vVols are present, run vVols migration as a parallel workstream — migrate VMs from vVols to VMFS-FC (or NVMe/FC) via Storage vMotion before the vVols termination date. SRDF replication relationships are preserved through the convergence.

Path C: PowerFlex + SDC upgrade + Brownfield Import. PowerFlex-backed estate. Sequence: upgrade SDC to the version supporting ESX 9 (SDC 5.1 or later, with partner-supported acceptance level on hosts), then upgrade vCenter and ESX hosts to v9, then bring vCenter under VCF via Brownfield Import. Note the special permission requirement: PowerFlex SDC integration into VCF requires explicit approval from VMware and Dell because the SDC modifies the ESX environment.

Path D: NVMe/TCP migration before VCF migration. For PowerFlex customers who want to move away from the SDC integration pattern, migrate to NVMe/TCP first (on the existing vSphere environment), then proceed with the VCF migration on the new protocol. This avoids the SDC-permission overhead and lands on a standards-based access protocol. Drew Tonnesen’s guidance: “if you are considering migration, definitely do so on ESXi 8 before going to ESXi 9. It will save you any hassle.”

Path E: Side-by-side with PowerStore Import. Standalone path leveraging PowerStore’s native import feature. Build a new VCF 9.1 instance on PowerStore. Use PowerStore’s agentless import to pull data from legacy storage (XtremIO, VMAX3, NetApp AFF-A, or existing PowerStore systems) directly into the new infrastructure. Migrate workloads on the consolidated storage. Useful when storage refresh is part of the project.

The vVols problem for Dell customers

Less acute than for NetApp or Pure, but real where it exists. vVols on PowerMax (and to a lesser extent PowerStore) is the typical Dell vVols footprint:

• Inventory: identify all vVols datastores and their VMs

• Target protocol: VMFS on FC (the standard PowerMax/PowerStore pattern), NVMe/FC, or NVMe/TCP

• Migration mechanism: Storage vMotion (Change storage only) via vCenter, or PowerStore/PowerMax native migration features

• Storage policies: rebuild SPBM policies for the target datastore type

• Replication: SRDF on PowerMax / native replication on PowerStore preserved through the migration

• Backup: PowerProtect / Data Domain backup workflows need validation on the new datastore type

Timeline: VCF 9.3 is the termination release for vVols per current industry signals. Customers should plan vVols migration to complete before VCF 9.3 adoption.

The PowerFlex SDC architecture transition

PowerFlex is unique in the Dell + VCF story because of the SDC. Architectural points:

• SDC is a kernel module installed on ESX hosts to consume PowerFlex storage

• Requires partner-supported acceptance level on the host: esxcli software acceptance set --level PartnerSupported

• Current SDC version for ESX 9.x: SDC 5.1 (covers all PowerFlex 5.x setups, including PowerFlex 5.0 and 5.1)

• PowerFlex VCF integration requires explicit approval from VMware and Dell — not a standard out-of-the-box path

• VMware has asked Dell to change how SDC integrates with ESXi — future SDC versions will conform to new VMware requirements

• NVMe/TCP is an alternative path that doesn’t require the SDC and is not affected by the integration changes

For HLD: if you’re a PowerFlex customer considering VCF 9.1, the strategic question is SDC vs NVMe/TCP. SDC gives you the existing operational model and Dell’s feature set. NVMe/TCP gives you a standards-based path without the partner-approval overhead. Many Dell customers are moving to NVMe/TCP as the long-term direction; the SDC continues but isn’t the strategic destination.

Replication and DR for Dell customers under VCF 9.1

PowerMax SRDF. SRDF in all its modes (Synchronous, Asynchronous, Adaptive Copy, SRDF/Metro) continues to work cleanly under VCF 9.x for VMFS-FC and NVMe/FC datastores. SRDF/Metro provides synchronous active-active for stretched cluster patterns.

PowerStore native replication. PowerStore’s synchronous and asynchronous replication continues to work for VCF 9.x. Metro replication (synchronous active-active) handles stretched cluster scenarios.

RecoverPoint (legacy). RecoverPoint for VMs continues to be supported but is being phased toward native array replication. New designs should prefer native array replication over RecoverPoint where possible.

PowerProtect Data Manager / Data Domain. Dell’s backup portfolio integrates with VCF 9.x via standard VADP and array-native snapshot APIs. PowerProtect Data Manager handles VM-aware backup; Data Domain provides the underlying protection storage. PowerProtect One (Dell’s newer offering) is generally available now.

Dell Cyber Detect. Coming in 2026: Dell Cyber Detect for PowerStore (Q3 2026) and Dell Cyber Detect for PowerMax (2H 2026). Provides ransomware detection at the storage layer, complementing VCF’s Advanced Cyber Compliance (Part 1) and vDefend (Part 5).

HLD design patterns for Dell + VCF 9.1

Pattern 1: PowerStore-all-FC. Management Domain on PowerStore VMFS-FC. Workload domains on PowerStore VMFS-FC. NVMe/TCP as supplemental for performance-sensitive workloads. PowerStore native replication for DR. Standard Dell + VCF design.

Pattern 2: PowerMax mission-critical. Management Domain on PowerMax VMFS-FC. Workload domains on PowerMax VMFS-FC. SRDF for synchronous or asynchronous replication. SRDF/Metro for stretched active-active. Suitable for the most demanding mission-critical workloads.

Pattern 3: PowerFlex with SDC. PowerFlex providing storage via SDC. ESX hosts with partner-supported acceptance level. SDC 5.1 installed. VCF 9.1 integration with VMware + Dell approval. Suitable for hyperscale and high-performance use cases where the operational depth of PowerFlex is worth the integration overhead.

Pattern 4: PowerFlex with NVMe/TCP. PowerFlex providing storage via NVMe/TCP. Standards-based access, no SDC required, no special permissions. The strategic direction Dell + VMware are pointing toward for PowerFlex + vSphere going forward.

Pattern 5: Dell Private Cloud. Dell-managed VCF 9.1 deployment (available June 2026), with disaggregated compute + storage scaling. Useful for customers who want VCF without operating the underlying stack themselves. DPC scales compute and storage independently — Dell’s positioning as up to 65% cheaper than HCI.

Common HLD/LLD anti-patterns

Treating PowerFlex SDC as a standard install. It requires partner-supported acceptance level on ESX hosts and explicit VMware + Dell approval for VCF integration. Plan it as a non-standard component.

Skipping the SDC + ESX coordination for PowerFlex upgrades. “If you did implement the VCF flat network model in 4.x or 5.x, you will need to work with Broadcom on the upgrade because it gets a bit complicated at what point to upgrade the SDC and reboot.” Migration sequencing matters.

Migrating PowerFlex on ESX 9 rather than ESX 8. Drew Tonnesen’s explicit guidance: migrate on ESX 8 first, then upgrade to ESX 9, rather than trying both steps in one go. Saves hassle.

Underestimating vVols migration timeline on PowerMax. VCF 9.3 is the termination release. Plan to be off vVols before adopting 9.3. Earlier is safer.

Ignoring PowerStore’s import feature. PowerStore can import data from XtremIO, VMAX3, NetApp AFF-A, or other systems. Useful for consolidation projects — but it’s an array-side operation, not a VCF operation. Plan it as a discrete workstream.

Mixing SRDF/Metro and VCF stretched cluster without validation. Both have their own RTT, link, and witness requirements. Validate both sets of constraints together in the LLD.

Assuming PowerProtect Data Manager works unchanged. VADP and array snapshot APIs evolve. Validate PowerProtect compatibility with VCF 9.1 and the target datastore type before cutover.

Waiting for Dell Private Cloud as a fix-everything option. DPC with VCF 9.1 lands June 2026. For customers needing to migrate before then, build directly on Dell hardware with VCF.

Decision framework: Dell customer paths

• PowerStore + VMFS-FC only, no vVols → Convergence (Path A)

• PowerMax + VMFS-FC, no vVols → Convergence (Path B without vVols workstream)

• PowerMax + vVols footprint → Convergence + parallel vVols migration (Path B with vVols workstream)

• PowerFlex with SDC, staying on SDC → SDC 5.1 upgrade then Brownfield Import (Path C)

• PowerFlex moving to NVMe/TCP → protocol migration on ESX 8, then VCF migration (Path D)

• Multi-array consolidation → PowerStore Import for storage consolidation alongside VCF migration (Path E)

• Want fully-managed Dell + VCF — wait for Dell Private Cloud + VCF 9.1 (June 2026)

• Mission-critical SRDF/Metro stretched → preserve SRDF/Metro, validate against VCF stretched cluster requirements

Closing

Dell + VCF 9.1 splits cleanly by product. PowerStore and PowerMax customers land in the standard VMFS-on-FC + supplemental NVMe pattern with replication portfolios preserved. PowerFlex customers face a more architectural decision: stay on SDC (with the partner-approval overhead) or move to NVMe/TCP as the standards-based path. Dell Private Cloud arriving in June 2026 adds a managed-offering option for customers who want the platform without building it themselves.

For my own design reviews with Dell customers, the questions I work through every time:

• Which Dell storage product — PowerStore, PowerMax, PowerFlex, or a mix?

• Source protocol: VMFS-FC, NVMe/FC, NVMe/TCP, vVols?

• PowerFlex customers: SDC path or NVMe/TCP path?

• If PowerFlex SDC — partner-supported acceptance level confirmed, VMware + Dell approval obtained?

• Replication topology: SRDF (PowerMax), native (PowerStore), PowerFlex native?

• Stretched cluster: SRDF/Metro, PowerStore Metro, or VCF’s own stretched cluster?

• PowerProtect Data Manager / Data Domain compatibility with VCF 9.1?

• vVols footprint on PowerMax: migration plan before VCF 9.3 adoption?

• Dell Cyber Detect adoption — layered with VCF’s ACC and vDefend?

• Dell Private Cloud appropriate for the customer’s operating model?

Get those answered and the Dell + VCF 9.1 HLD writes itself.

That closes Part 12 — the Dell-specific overlay on the VCF 9.1 migration journey. Part 13 covers the same migration journey for HPE customers.

Sources

• VCF 9.1 and Dell Storage (Drew Tonnesen, Dell)

• VCF 9 / VVF 9 — Dell PowerFlex, PowerMax, and PowerStore (Drew Tonnesen, Dell)

• VCF 9 PowerMax Implementation Guide (Drew Tonnesen, Dell)

• VCF 9.0 and PowerFlex 5.0 (Drew Tonnesen, Dell)

• Dell PowerFlex 5 and vSphere 9 (Drew Tonnesen, Dell)

• Dell PowerStore Importing External Storage to PowerStore Guide (Dell)

• PowerStore Gets Performance and Capacity Upgrades — Dell Private Cloud (Blocks & Files)