What Would It Actually Take to Replace VCD's Abstraction Layer?

Cloud Architecture - MSP Strategy
VMware Cloud Director MSP Strategy Platform9 CloudStack Nutanix Open Question
The question MSPs keep asking is the wrong question. "What replaces VCD?" assumes VCD is a product. It is not - it is a role. Three specific jobs that, taken together, define what it means to run a multi-tenant private cloud service. Until you define those jobs precisely, every answer to the replacement question is going to miss at least one of them. And so far, every platform in the market is missing at least one of them.
🧩

1. The Three Jobs VCD Actually Does

Most replacement conversations focus on the hypervisor. That is the wrong layer. VCD does not compete with vSphere at the compute layer - it sits above it. The hypervisor is interchangeable. The abstraction layer is not. Strip out the hypervisor discussion entirely and VCD's real value is in three distinct jobs it performs simultaneously, and which every MSP who has run it takes for granted.

Job 1: Tenant Isolation and Resource Delegation

VCD gives MSPs a complete, hierarchical multi-tenant model. Provider VDCs pool the raw compute. Organizations sit above that and receive allocated slices. Within each organization, Org VDCs define the resource envelope. Below that, the three-tier model adds Sub-Provider organizations that can resell capacity to their own tenants without visibility into the underlying infrastructure. Every boundary in this hierarchy enforces network isolation, resource quotas, and RBAC natively. The tenant in Org A cannot see, affect, or consume resources belonging to Org B. This is not achieved through operational discipline - it is enforced by the platform.

Most platforms can do tenant isolation at some level. What is harder to replicate is the depth of the delegation model - specifically the sub-provider tier that lets MSPs build a reseller channel without exposing their infrastructure. That construct does not exist in most alternatives out of the box.

Job 2: Service Catalog and Billing Surface

VCD gives MSPs a billable service surface. Catalogs contain vApp templates and media. Tenants deploy from those catalogs. Every resource consumed - vCPU, vRAM, storage, network egress - flows into Usage Meter, which aggregates it against the VCSP licensing model and produces the monthly report that drives the Broadcom Commerce Portal invoice. White-labeling lets the MSP brand the portal. The entire commercial wrapper - tenant self-service, resource metering, billing integration - is built in.

This is the job that most VCD discussions underweight. The service catalog and billing surface is not a nice-to-have. It is the commercial engine of the MSP's business. A replacement platform that forces you to build your own billing integration from scratch, or bolt on a third-party commercial layer, is not replacing VCD - it is giving you the infrastructure layer and asking you to rebuild the business layer yourself.

Job 3: Third-Party Integration Surface

VCD exposes a well-defined API that a decade of the ecosystem has built against. Veeam Cloud Connect, Zerto Cloud Manager, and a long tail of monitoring, compliance, and automation tooling all integrate against that API surface. These are not generic tools that happen to run on VMs hosted on VCD infrastructure - they are tools with deep, VCD-specific integration that gives MSPs a complete commercial model layered on top of the compute service. Remove VCD and you do not just lose the platform - you lose the integration surface that makes those commercial models work.

When the replacement question comes up, nobody talks about this job. That is a mistake, because this is the one where the gap in the market is widest and the switching cost is highest. Section 2 breaks down exactly why.

🔗

2. The Ecosystem Integration Problem

Veeam Cloud Connect is not just backup software that works on VMs. The VCD integration handles multi-tenant namespace isolation so that one tenant's backup jobs, repositories, and restore operations are completely invisible to every other tenant on the same platform. VSPC sits on top of that and gives MSPs a single pane of glass across all tenant backup operations with per-tenant billing data. The self-service portal lets tenants manage their own backups without ever touching the MSP's administrative layer. That entire model - the isolation, the self-service, the billing integration - is built on top of VCD's API. Remove VCD and you remove the foundation that model runs on.

Zerto's story is similar. The ZVM-to-VCD integration handles tenant-level failover operations through VCD's Org VDC model. Tenants see their protected VMs fail over into their Org VDC. The multi-tenant isolation means a failover operation for one tenant cannot affect another. Strip out VCD and you have ZVMs that can still replicate VMs, but the tenant-isolated failover model that makes it a sellable DRaaS product no longer works the same way.

What Veeam Said Publicly When MSPs Asked About Platform Alternatives
When displaced VCSPs started asking Veeam directly about Platform9 support, a Veeam product manager responded on their public R&D forum thread with language that is worth reading carefully: most VCSP partners intend to remain on VCD for an extended period, with no one planning to rush migration to VCF Automation. And on the question of alternative platforms - Veeam heard very different opinions from different VCSPs about which alternative is compelling, and said they would need to see consolidation in those opinions as well as meaningful market share before deciding where to expand platform support next. That is a considered business decision, not an evasion. But it defines the problem precisely: the ecosystem is waiting for the market to pick a winner before it commits integration resources. And the market is waiting for ecosystem support before it migrates. Nobody blinks first.

Zerto, which HPE acquired and which now runs as a distinct product within the HPE portfolio, has its own integration calculus. The VCD integration is mature and well-maintained. Zerto on Nutanix AHV and Hyper-V exist. But the VCD-specific multi-tenant model - where the MSP can deliver DRaaS through a tenant portal with VCD-native failover - has no direct equivalent on any other platform today.

The replacement question is therefore actually four questions stacked on top of each other. What replaces the compute and isolation layer? What replaces the service catalog and billing surface? What replaces the Veeam CC multi-tenant BaaS model? What replaces the Zerto multi-tenant DRaaS model? Any platform that cannot answer all four is a partial replacement at best.

📊

3. The Contenders - Honest Assessment

The following assessments are against the three jobs defined in Section 1. Not against the hypervisor layer. Not against feature checklists. Against the specific role VCD plays in a functioning MSP business.

Platform9 Private Cloud Director
OpenStack-based management plane, SaaS or self-hosted. KVM hypervisor.
Closest Contender
Job 1 - Tenant Isolation
YES - domains, tenants, per-tenant quotas, RBAC, network isolation. Built for MSPs from the start, not adapted after the fact.
Job 2 - Service Catalog / Billing
PARTIAL - Terraform-based application catalog. No built-in billing equivalent to Usage Meter. Third-party integration required.
Veeam Integration
BETA - Veeam Universal Hypervisor API via oVirt adapter. Beta as of February 2026. Standard backup only. No Cloud Connect multi-tenant model, no VSPC equivalent yet.
Zerto Integration
NOT ANNOUNCED - No Zerto integration for PCD as of March 2026.

Platform9 is the most serious contender in the market today for MSPs coming off VCD. The multi-tenant management plane is built for service provider operations from the ground up - not a single-tenant product with a multi-tenancy bolt-on. The vJailbreak migration tool for converting VMware clusters in-place is a real differentiator for MSPs who cannot afford a forklift migration. The Veeam integration entering beta is meaningful progress. The gap is the commercial layer: the Veeam Cloud Connect multi-tenant BaaS model and the VSPC billing integration do not have equivalents yet, and there is no Zerto path. An MSP whose revenue model depends on BaaS and DRaaS services cannot fully replace VCD with PCD today. That gap may close - Platform9 is moving fast - but it is not closed yet.

Apache CloudStack
Open source IaaS platform. Hypervisor agnostic: KVM, XCP-ng, VMware, Hyper-V.
Battle-Tested, Assembly Required
Job 1 - Tenant Isolation
YES - accounts, domains, zones, VPC networking. Fifteen years of MSP production usage. The multi-tenancy model is mature.
Job 2 - Service Catalog / Billing
OVERLAY REQUIRED - billing and self-service portals via Apiculus, WHMCS, or similar. The bones are there via CloudStack's usage API, but the commercial layer is not built in.
Veeam Integration
AGENT ONLY - Veeam agent-based backup works for CloudStack VMs. No hypervisor-level integration, no Cloud Connect multi-tenant model, no VSPC path.
Zerto Integration
NOT SUPPORTED

CloudStack is the answer for MSPs building a public IaaS business on open infrastructure - particularly those running KVM or XCP-ng and willing to compose their own commercial layer. Several MSPs and telcos have been running it in production for years with real scale. The integration maturity gap with Veeam and Zerto is the same story as Platform9 - agent-based backup exists, but the multi-tenant BaaS commercial model that VCD enables through Veeam CC has no equivalent. The difference from Platform9 is that CloudStack requires more assembly work to get to a finished MSP product. You are not buying a product - you are building one from components, some open source, some commercial overlays.

Nutanix Cloud Platform
AHV hypervisor, Prism Central management, Flow Virtual Networking, NCM Self-Service.
Capabilities Without Packaging
Job 1 - Tenant Isolation
PARTIAL - Flow Virtual Networking provides VPC-style tenant network isolation. RBAC exists. But there is no discrete MSP control plane layer equivalent to VCD's Org/Sub-Provider model. Tenants interact with Prism, not a white-labeled service portal.
Job 2 - Service Catalog / Billing
PARTIAL - NCM Self-Service provides Terraform-based blueprints and self-service provisioning. Nutanix Central has chargeback and metering. No VCSP Usage Meter equivalent, no Commerce Portal billing path.
Veeam Integration
HYPERVISOR LEVEL - Veeam v13 added full AHV backup support. Standard enterprise backup works well. No Cloud Connect multi-tenant model. The Elevate Service Provider program exists as a commercial framework, but it is a partner program, not a product feature.
Zerto Integration
HYPERVISOR LEVEL - Zerto runs on AHV. VM-level replication works. No VCD-equivalent multi-tenant DRaaS model with tenant portal isolation.

Nutanix has all the raw ingredients - CDP-like replication, multi-tenant constructs, the underlying compute fabric - but none of it is packaged as a discrete MSP control plane. VCD's value was never the hypervisor, it was the abstraction layer that let MSPs build a service catalog, delegate resources to tenants, and bill against it. Nutanix has not shipped that equivalent yet. MSPs running Nutanix today are running Prism Central and building their own service wrapper around it. That is a workable model for large, sophisticated operators who want to own the entire stack. It is not a drop-in VCD replacement for the mid-market MSP who needs to be selling services, not building platform infrastructure. The opportunity for Nutanix is real, but the packaging is not there yet.

OpenStack
Open source cloud platform. Technically answers every requirement. Operationally demanding.
Technically Complete, Operationally Brutal
Job 1 - Tenant Isolation
YES - projects, domains, Keystone auth, Neutron networking with full tenant isolation. The multi-tenancy model is mature and complete.
Job 2 - Service Catalog / Billing
ASSEMBLY REQUIRED - Heat for orchestration, Cloudkitty for billing metering, Horizon for UI. All the pieces exist. None of them ship assembled.
Veeam Integration
AGENT ONLY - same story as CloudStack. No Cloud Connect model.
Zerto Integration
NOT SUPPORTED

OpenStack technically answers every question in this article. The multi-tenancy model is complete. The API surface is rich. The scale ceiling is high. The MSPs running OpenStack at scale - OVH, Rackspace, a handful of large telcos - have built what are essentially internal product companies to operate it. That operational burden is the real cost of OpenStack, and it is not something that scales down to a 20-person MSP. The MSPs who have tried and failed at OpenStack generally did not fail because OpenStack could not do the job - they failed because operating OpenStack continuously over years requires engineering investment that does not make economic sense at mid-market MSP scale.

Azure Local (formerly Azure Stack HCI)
Microsoft on-prem cloud extension. Deep Azure Arc integration. Rebranded from Azure Stack HCI in November 2024.
Right Answer for the Wrong Problem
Job 1 - Tenant Isolation
YES - Arc-enabled multi-tenant management. Full isolation at the Azure subscription level.
Job 2 - Service Catalog / Billing
YES - Azure Marketplace, Arc catalog, Azure billing surface. Best commercial layer of any platform here - if you are selling Azure-adjacent services.
Veeam Integration
YES - Full Veeam integration. Best non-VCD Veeam story in this list.
Zerto Integration
YES - Full Zerto integration for Hyper-V and Azure Local.

Azure Local scores the best here on paper, and that is precisely why it is a trap for some MSPs. The integration depth with Veeam and Zerto is real. The multi-tenant model works. The commercial layer is the richest of any platform in this list. But the model it puts you in is fundamentally different from what VCD enabled. VCD let MSPs operate a platform-independent IaaS business. Azure Local puts you in a position where your service is an extension of Microsoft's cloud, billed through Azure meters, governed by Microsoft's partner program. If your business model is "I am an Azure extension provider," this is the right platform. If your business model is "I run a sovereign, platform-independent private cloud service," this answers a different question than the one you asked.

The Matrix

Platform Tenant Isolation Service Catalog / Billing Veeam CC Model Zerto MSP Model
VMware Cloud Director Full - Org / Sub-Provider model Full - Usage Meter / Commerce Portal Full - VSPC / self-service portal Full - ZCM / tenant portal
Platform9 PCD Full - domain / tenant model Partial - no built-in billing Beta - backup only, no CC model Not available
Apache CloudStack Full - domain / account model Partial - overlay required Agent only - no CC model Not supported
Nutanix Partial - Flow VPC, no MSP layer Partial - NCM, no VCSP billing Hypervisor level - no CC model Hypervisor level - no MSP model
OpenStack Full - project / domain model Partial - assembly required Agent only Not supported
Azure Local Full - Arc / subscription model Full - Azure billing surface Full - but Microsoft ecosystem only Full - but Microsoft ecosystem only
🚧

4. The Gap Nobody Has Closed

Look at that matrix and the gap is clear. Every platform in the list resolves the compute and isolation layer to some degree. Every platform fails, partially or completely, on the commercial layer - specifically the Veeam Cloud Connect multi-tenant BaaS model and the Zerto multi-tenant DRaaS model. Azure Local is the exception, but it solves the problem by locking you into a different ecosystem rather than by replicating what VCD enabled.

The Veeam forum response quoted in Section 2 is the clearest articulation of why this gap persists. Veeam is not withholding investment out of stubbornness - they are making a rational business decision. Building and maintaining a deep multi-tenant Cloud Connect integration against a new platform is significant engineering work. They will do it when enough MSPs have migrated to a specific platform that the economics justify it. The MSPs who want to migrate are waiting for that integration before they commit. The market is locked in a standoff where nobody moves first.

This is not a new dynamic in the infrastructure industry - it is the classic platform adoption problem. But it is particularly acute here because the integration that MSPs need is not a simple API connection. The Veeam CC multi-tenant model is architecturally deep. Replicating it on a new platform requires Veeam to build a new integration from a near-zero starting point, not port an existing one.

The Question Nobody Is Asking About VCF Automation
Broadcom's own successor to VCD within the VCF 9.0 stack is VCF Automation (formerly Aria Automation). Veeam stated publicly that Cloud Connect support for VCF Automation is not yet in place and that they have time before mass migration starts. This means that even the platform Broadcom intends as VCD's successor does not yet have full Veeam CC support. The Broadcom-to-Broadcom migration path is no more complete than the migration paths to any third-party alternative. If you are staying on VCD because you are waiting for a clear successor with full ecosystem support, you are waiting for something that does not yet exist in any direction.
🔭

5. What Comes Next

The market will resolve this. The question is how and on what timeline.

Platform9 is the most likely platform to close the gap first, and the trajectory matters. They have the multi-tenant management plane. They have the Veeam backup beta going. They have an active partnership and migration story. If the Veeam integration reaches GA, adds a Cloud Connect equivalent for tenant-isolated BaaS, and a Zerto integration is announced, Platform9 moves from "interesting alternative" to "viable VCD successor" for most MSPs. None of those things have happened yet as of March 2026, but the direction is right.

CloudStack closes the gap differently - not through a single platform vendor but through ecosystem consolidation. If Veeam builds Cloud Connect support against CloudStack's hypervisor API, the large number of MSPs already running CloudStack in production suddenly have a complete story. This is possible but requires Veeam to make the investment decision, which brings us back to the standoff problem.

Nutanix has the market presence and the engineering resources to build a discrete MSP control plane if they decide that is the product direction. The raw ingredients are there. The packaging decision is a strategic one, not a technical one. An MSP-facing product layer built on top of Prism Central and Flow VPC - with a VCSP-equivalent billing path and a Veeam CC integration - would be a real competitor. Whether Nutanix builds that or positions the platform as infrastructure that MSPs wrap themselves is a question that their roadmap will answer over the next 18 months.

The MSP who wins in this transition is not necessarily the one who picks the right platform earliest. It is the one who correctly reads which platform reaches ecosystem integration completeness first and positions their service catalog around it before their competitors do. That requires watching Platform9's Veeam integration go GA, watching Veeam's forum for announcements on platform expansion, and watching Nutanix's product direction through .NEXT 2026. The signal will be clear when it comes. Right now, the honest answer to "what replaces VCD?" is: nothing does, completely, yet.


Read more