A client asked me in July whether they should start planning their move from VCF 5.x to 9. Reasonable question, the kind I'd expect to get more of as this series runs. The honest answer took longer to land on than either of us expected, and it wasn't really about the technology.
Their licensing came up almost immediately. They're still on perpetual VCF licenses, and Broadcom's own FAQ is blunt about what that means: perpetual license holders can't upgrade straight into VCF 9. That's a conversation with a reseller that has to happen before anything technical gets scoped — not a hidden gotcha exactly, just a line buried in a document nobody reads until they're already asking the question out loud.
There's No Deadline Anymore, Which Changes the Question
Broadcom has said outright there's no mandatory cutoff for moving off VCF 5.x onto 9. No forced upgrade window, no hard deadline pushing the decision. That's a real departure from how VMware upgrades have historically felt, where falling a version or two behind started generating urgency fast, whether or not the urgency was actually earned.
VCF 5.x sits under the current Enterprise HCI support policy — four years of General Support, followed by a year of Technical Guidance once that closes out. Depending on where an environment actually sits inside that window, "should we move now" and "should we move eventually" are two different questions with two different answers. Treating them as the same question is how a rushed migration happens that nobody actually needed to rush.
Without a forced deadline, readiness stops being a compliance question and becomes a business one. That's a harder conversation to have well, not an easier one to skip.
VCF 9 also introduced a different support cadence going forward — a longer major-version support window paired with minor releases roughly every nine months instead of the older pattern. Worth knowing before deciding when to move, because it changes what "falling behind" even means once you're on 9.
What's Actually Gone, Not Just Renamed
This is where I'd start the technical side of the assessment, honestly, more than the licensing conversation. VCF 9.0 removed Integrated Windows Authentication for vCenter SSO entirely. Not deprecated with a warning label — gone. If anything in the environment still leans on IWA for authentication, that's a dependency that needs to be dealt with before migration planning starts, not discovered halfway through it when someone can't log in the way they used to.
IWA is the one I run into most often because it's easy to forget it's even in use — it tends to be configured once, years earlier, by someone who's since moved on. It's not the only removed feature in 9.0, and checking the full deprecated-and-removed list against the current environment is a five-minute task that saves a much longer conversation later.
Cloud Builder Doesn't Exist Anymore Either
Anyone who's deployed VCF before knows Cloud Builder as the appliance that kicks off a bring-up. In VCF 9, that role belongs to the VCF Installer instead — a different appliance, with automated prechecks built in that Cloud Builder never had. It's not a drop-in replacement conceptually, even though the outcome looks similar. Anyone planning a migration who's mentally picturing the old Cloud Builder workflow is planning against a tool that's been retired.
Confirm the licensing model. Perpetual license holders need a subscription conversion conversation with their reseller before anything technical starts.
Check where the environment sits in the VCF 5.x support window — early, mid, or approaching the end of General Support changes the urgency, not the eventual answer.
Cross-check the environment against VCF 9.0's deprecated and removed features list, starting with authentication dependencies like IWA.
Confirm the target hardware and BOM are supported for VCF 9 — don't assume current ReadyNode compatibility carries forward unchanged.
Understand that Cloud Builder is retired — migration planning has to account for the VCF Installer workflow, not the old one.
None of this is an argument against moving to VCF 9. The platform itself is a genuine step forward, and nothing above changes that. It's an argument for treating "when" as a real question with a real answer specific to the environment in front of you, instead of a default that just happens whenever the next vendor announcement creates enough noise to prompt it.