Introducing billing for ephemeral nodes and tagged resources behind an opaque "Contact Sales" wall doesn't feel like generosity, it feels like a classic bait-and-switch.
We architected our infrastructure around Tailscale (under their "now legacy" Premium plan) under the reasonable assumption that these specific usage patterns wouldn't suddenly become cost centers. For context, we run on-prem Kubernetes with Flannel as CNI in host-gw mode, using Tailscale purely as the underlying transport. Because of this architecture, every Kubernetes node acts as a subnet router, which now neatly falls into their newly monetized "tagged resource" bucket.
Because of this pricing change, we're now looking at a one-year ticking clock. Our options are to either walk into an enterprise sales negotiation at a severe information asymmetry disadvantage to keep our current architecture, or rip out our networking layer entirely. I've already added an "Evaluate Netbird" to our team's backlog.
So it's deeply disappointing, but perhaps we should have seen it coming. I already perceive this as the standard lifecycle of a VC-backed HN darling: build immense goodwill with developer-friendly terms, embed yourself as a deep infrastructure dependency, and then aggressively squeeze the margins once the lock-in is established.