TL;DR — Key Takeaways

  • VDDK is gone, but the bigger headache remains: Broadcom’s withdrawal disrupts migration tools, while undocumented network policies threaten to derail entire VMware exit plans.

  • Moving VMs is easy. Rebuilding security isn’t: NSX firewall rules, security groups and application dependencies don’t translate neatly between platforms, creating costly migration bottlenecks.

  • Policy first, data second: FluidCloud’s vHOP aims to map infrastructure, translate firewall policies and generate Terraform before workloads move, reducing security risks and renewal pressure.

When Broadcom took down the public download pages for its Virtual Disk Development Kit (VDDK) in late August, a lot of VMware exit plans broke. Many agentless migration tools used the library to read virtual disks. Microsoft Azure Migrate, Red Hat’s Migration Toolkit for Virtualization and Nutanix Move can all require it for certain workflows.

Broadcom Customer Care has told users that VDDK is “no longer available for use or download.” The company says it still maintains the APIs and SDKs that authorized technology alliance partners use to build backup and recovery products.

For teams halfway through an exit, that means re-planning around in-guest agents or cold copies. And many of them are doing it with a VMware renewal date on the calendar.

FluidCloud CEO and co-founder Sharad Kumar thinks the industry is watching the wrong problem. Losing VDDK made disk copying harder, he says. But disk copying rarely stalled these projects. Kumar says what pushes migrations past renewal dates is “the firewall rules, the security groups and the dependencies nobody wrote down.”

Mitch Ashley, vice president and practice lead for CIO & Technology Buyers and Software Lifecycle Engineering at The Futurum Group, sees it the same way. “A VMware exit is a policy migration first and a data migration second,” he says. “A firewall rule base holds years of security decisions, and many were never written down. Someone has to re-decide what each rule was for.”

The Translation Gap

Kumar’s argument is simple. Migration tools move VMs. Then they leave the network to you.

“There is no translation layer, so whoever lands the VMs rebuilds the networking and the firewall by hand, and NSX does not resemble what they are rebuilding it in,” Kumar says.

The mismatches are basic, and there are many. NSX rules attach to groups and VMs, and the target may have no equivalent groups. NSX evaluates rules in priority order, and some clouds don’t. NSX can deny traffic. AWS security groups can only allow it.

Kumar uses a simple example. An NSX rule denies port 22 from any address and allows everything else. To reproduce that on AWS, you need two allow rules from 0.0.0.0/0, one for ports 0 to 21 and one for ports 23 to 65535.

“One rule became two, the logic inverted, and that was the simplest case,” he says. “Multiply it by priority ordering, nested groups and a few thousand policies, and the manual version is where the errors creep in.”

There’s a people problem, too. “The engineer who knows NSX does not know the target cloud, the engineer who knows the cloud has never read an NSX rule base, and neither knows how the customer’s applications actually use the network,” Kumar says.

So the work turns into consulting hours. Kumar says one FluidCloud customer with about 1,500 policies was looking at four months to convert them by hand. He says the conversion took minutes with FluidCloud’s platform.

Plan the Network Before Moving the Data

FluidCloud’s answer is vHOP, short for VM Hopping Engine. It launched October 27 as a VMware exit package that maps the estate, translates NSX firewall policy, and generates Terraform for the target cloud. It does all of that without VDDK and before any workload data moves.

The engineer-led version takes three days and covers five steps:

  • Agentless, read-only discovery of VMs, networks, NSX groups and firewall policies through the vCenter and NSX Manager APIs.

  • A review of which migration, backup and disaster recovery tools in the current toolchain depend on VDDK.

  • A firewall migration plan that translates NSX rules for the target cloud and flags where target quotas will be hit.

  • Terraform modules for the destination cloud.

  • A recommended data path for each workload group. That can be FluidCloud’s own migration engine, which doesn’t use VDDK, or an existing tool such as AWS Transform MGN.

A self-service assessment returns a plan in minutes. Supported targets are AWS, Microsoft Azure, Google Cloud, Oracle Cloud Infrastructure and Nutanix. vHOP is available in the FluidCloud platform and through AWS Marketplace.

Order matters here, and Kumar says many teams get it wrong. They assume security policy can wait until after the move.

“They cannot. Rules have to exist before the VMs land, because they attach to the VM at creation,” he says.

Because vHOP’s output is code, teams can check it before anything deploys. FluidCloud runs a security analysis on the source estate. It runs the same analysis on the generated Terraform and compares the two. Kumar calls the result “a posture report for a cloud environment that does not exist yet.” He says that, to the company’s knowledge, it’s a first. Where the target can’t express a source rule, the team sees it in the code review before deployment.

Same Idea, Different Problem

The same approach shows up in FluidCloud’s resilience work. On November 10, the company announced it had earned the AWS Resilience Software Competency in the Resilience Recovery category. Its platform discovers a live AWS environment and turns it into deployment-ready architecture code covering Amazon VPC, AWS IAM, Elastic Load Balancing and Amazon RDS. It can then rebuild a working environment in another AWS Region or account.

Kumar calls the gap between backed-up data and working infrastructure “pseudo-resiliency.” You get your data back, but the business stays down while someone rebuilds networking and IAM by hand.

Bay Area Rapid Transit (BART) is a customer. “In public transit, operational downtime is not just an IT inconvenience; it directly impacts regional mobility,” says Ravi Misra, BART’s CIO and assistant general manager of technology.

Whether it’s a VMware exit or a regional outage, the problem is the same. The data is the easy part. The undocumented configuration around it is what slows recovery.

Ashley says that’s why the policy has to move the same way the infrastructure does. “Security policy has to move as code a team can review before cutover, because hand-rebuilt rules fail quietly,” he says. “Watch whether teams can prove the target enforces the same policy as the source before the renewal date forces the move.”

What to Do Now

The clock is running for VMware customers. General support for vSphere 8, the last perpetual version, ends in October 2027. Teams planning an exit should inventory their NSX rule base before choosing a data mover. Find out which tools in the stack depend on VDDK. And don’t let the network become something you rebuild after the move.

The VDDK decision drew the headlines. The firewall rules are what will decide when these projects actually finish.

Frequently Asked Questions

Why is Broadcom's VDDK withdrawal a problem for VMware migrations?
Many agentless migration tools depend on VDDK to read VMware virtual disks. Without public access to the library, organizations may need alternative migration methods, including in-guest agents or cold copies.
Why are firewall policies a bigger challenge than moving virtual machines?
VMware NSX and destination cloud platforms handle security rules differently. Recreating thousands of rules manually can introduce configuration errors, weaken security and push migrations beyond critical renewal deadlines.
How does FluidCloud's vHOP simplify VMware exits?
vHOP discovers existing infrastructure, translates NSX firewall policies, identifies compatibility issues and generates reviewable Terraform code. This allows teams to validate network security before migrating workloads, without relying on VDDK.