DEV Community

NTCTech
NTCTech

Posted on Originally published at rack2cloud.com

"Decide Later" Is Not an Architecture Strategy

Azure VMware Solution is putting a useful architectural lesson on display: postponing a platform decision can feel like preserving optionality, right up until someone else puts an expiration date on it.

Microsoft confirmed this month that AVS's license-included model is going away on a fixed schedule — new sales already stopped over a year ago, and existing customers now have concrete dates to convert to a bring-your-own-license model. The specifics don't matter much here (they're covered in full on Rack2Cloud if you want them). What matters is the shape of the mistake it's exposing, because it isn't a VMware mistake. It's a pattern that shows up anywhere a platform decision gets deferred long enough to stop feeling like a decision at all.

The Fourth Option Isn't a Real One

Most teams sitting on an aging platform believe they have four choices: keep it, migrate off it, modernize around it, or decide later once things are clearer. The first three are real positions. Each has a cost model, a timeline, trade-offs you could actually defend to someone asking hard questions. The fourth isn't a position — it's the absence of one, dressed up to look like patience.

Four options, three of them real: Stay, Migrate, and Modernize shown as solid filled boxes; Decide Later shown as an empty dashed outline

The tell is how each one behaves under pressure. A real decision degrades gracefully: you re-run the analysis, adjust the timeline, maybe reverse course, but there's an actual call sitting underneath the discomfort. Deferral doesn't degrade. It just sits there, quietly converting elapsed time into constraint, until something external forces the question nobody internally ever did.

AVS Is the Case Study, Not the Point

For AVS specifically: organizations still on the license-included model now have hard dates — different ones depending on whether they're on PayGo or Reserved Instances, which is its own small lesson (there usually isn't one universal deadline, there are several, and they land on different teams differently). Two years of "we'll revisit this" didn't produce a stronger negotiating position for anyone. It produced a compressed timeline and a narrower option set, because the fourth option was never actually buying time. It was borrowing it, against a bill nobody had dated yet.

Swap out AVS for whatever's sitting in your own backlog under "revisit eventually" — a staging environment nobody's decommissioned, a hybrid arrangement that outlived the reason it existed, a vendor contract nobody re-reads. Same mechanism. A platform decision doesn't get made in one moment where someone weighs the options and picks one. It gets made — or fails to get made — in a long accumulation of "not yet," each individually reasonable, none of them adding up to an actual decision, until an external party sets a date that makes the accumulation visible all at once.

Deliberate Coexistence vs. Accidental Coexistence

There's a real distinction worth drawing here, because "we haven't decided yet" and "we've deliberately chosen to run both for now" can look identical from the outside and mean completely different things architecturally.

Deliberate coexistence has an owner, a review date, and a stated condition for when it ends. Accidental coexistence just has a start date nobody remembers and no mechanism that would ever force a revisit. The first is a strategy. The second is what deferral looks like once enough time has passed for people to stop noticing it's happening.

Any hyperscaler-hosted arrangement — not just AVS — runs on some vendor's current assumption about how long today's terms hold. None of them come with a guarantee that assumption survives eighteen months. The teams that treat these as provisional, worth a standing review cadence rather than a set-and-forget decision, experience the next version of this as a scheduled check-in. Everyone else experiences it as a deadline.


Microsoft didn't create the pressure AVS customers are feeling right now. It just picked the date the pressure became visible. Worth asking of your own stack: what's the platform decision you've been quietly not making — and whose calendar is going to be the one that finally forces it?


Full architectural treatment, with the complete AVS licensing timeline, at rack2cloud.com.

Top comments (0)