When this feature was announced in 2018, I wrote:
<sarcasm> It’s like linked servers, but since they don’t perform well, we need to scale out across containers. </sarcasm>
I just didn’t get it, and I continued:
I like that Microsoft is making a risky bet, planting a flag where nobody else is, saying, “We’re going to be at the center of the new modern data warehouse.” What they’re proposing is hard work – we all know first-hand the terrible performance and security complexities of running linked server queries, and this is next-level-harder. It’s going to take a lot of development investments to make this work well
The thing with bets, as the great strategist K. R. Rogers wrote, is you gotta know when to fold ’em, know when to walk away, and know when to run.
Today, Big Data Clusters died in its sleep.
Back then, I said I liked that Microsoft was making risky bets. They’ve proven time and again that they’re willing to continue to do that with SQL Server, throwing all kinds of crazy features against the wall. Big Data Clusters in Kubernetes. Machine Learning Services on Linux. Calling Java code from T-SQL. I look at a lot of these bets and think, “Uh, I don’t think you can win a poker game with two threes, a Subway discount card, and a happy birthday card from Grandma. That’s not how this works.”
The great part about these utterly wacko bets is that they don’t break the database engine. Microsoft SQL Server is still a robust, powerful persistence layer with a lot of good features. Microsoft hasn’t made dumb moves like saying, “We’re going to replace T-SQL with interpretive dance,” or “We’re not going to run queries if they don’t end in a semicolon.” They’ve figured out that they have to keep the core engine working better and better with each release, too.
If you ask me, five nines is a better hand.
There’s a lesson in here when you gamble on your own career. I beat this drum a lot, but I’m gonna say it again: you only have so many hours in the day to learn new skills. Make sure every hour counts.