MicrosoftThree years and CDC in ADF is still in Preview. Meanwhile, we were pushed to move to Synapse, which was essentially its own flavor of ADF, only to arrive at 3 FabCons later with Microsoft now marketing to move off Synapse entirely and onto Fabric. At some point this pattern stops being a roadmap and starts being a revolving door.
Time to market matters, but shipping quality matters more. Preview tags used to signal "nearly there." Now they signal "indefinitely parked." The problem is that marketing does not communicate that nuance. CIOs see a Microsoft preview, greenlight it, and have teams build production pipelines around it with an expectation of GA within a year. Those teams are still waiting.
And then there is the Fabric capacity model itself, which deserves its own conversation. Microsoft's own documentation confirms that when a Fabric capacity is overutilized beyond 24 hours of compute unit allocation, all activities are rejected until the overage is reduced. Every workload shares the same capacity pool: Power BI reports, data pipelines, notebooks, Spark jobs, everything. One poorly designed pipeline left running overnight can consume the entire capacity and freeze your reporting, your data movement, and your operations simultaneously. Real world cases of this happening are already well documented in the community.
What data infrastructure company sells a platform where a single bad pipeline can take down your entire analytics estate? That is not enterprise grade. That is a capacity model built for demos.
The community deserves transparency over hype. Are these features being actively developed, or are we just being steered toward the next product cycle before the current one is finished?
