Where database blog posts get flame-broiled to perfection
I happened upon this... article, forwarded to me by a graduate student in a moment of what I can only assume was profound intellectual despair. The title alone, a frantic concatenation of buzzwords, is a symphony of category errors. It reads less like a technical abstract and more like a desperate plea for venture capital. Still, one must occasionally survey the wilderness to appreciate the garden. Here are a few... observations.
They prattle on about real-time data as if physics and the CAP theorem were mere suggestions offered by a timid subcommittee. In their world, one can apparently have one's cake, eat it too, and have it delivered instantaneously across three availability zones with no consistency trade-offs. It's miraculous. They have solved distributed computing, and all it took was ignoring the last forty years of it. I suppose when your goal is a flashy dashboard, the "C" in ACID is merely the first letter in "Close enough."
The obsession with "rich, generative analytics UIs" is a classic stratagem: dazzle them with glistening charts so they don't notice the rampant data skew and double-counting happening just beneath the surface. 'But look, professor, the bars animate!' Yes, my dear boy, so does a lava lamp, but I wouldn't use one to perform relational calculus. This is the art of the sophisticated lie, dressing up a probable data swamp in the garb of a spring-fed intellectual oasis.
One must assume the authors view the foundational principles of database systems as a quaint historical footnote, perhaps filed somewhere between phrenology and the belief in a geocentric universe. The entire premiseâshoveling data into a high-velocity analytical engine and calling it a dayâsuggests a complete and utter disregard for transactional integrity. Atomicity? Isolation? These are the concerns of dusty old academics, not disruptors. Clearly, they've never read Stonebraker's seminal work on the trade-offs between OLTP and OLAP systems. It's all there, children. In the primary sources.
And the very foundation! The casual observer might think these systems are databases, but they fail to adhere to even the most basic of Codd's rules. What of the information rule? The guaranteed access rule? It seems the only rule they follow is that if you put a sufficiently sleek API in front of a glorified, indexed log file, someone in marketing will call it a revolution.
Learn how to create... UIs backed by real-time data A more honest title would be: 'Learn how to generate plausible-looking fictions from a chaotic firehose of information.'
A charming, if deeply misguided, piece of ephemera. I shall not be reading this blog again.