O’Reilly Report Shows Relational DB Momentum, No AI Slowdown
O’Reilly has released its Technology Trends for 2022 report, highlighting trends in its data spanning technologies from cloud and AI to databases and the metaverse. Among its conclusions: a bright and continuing future for AI—particularly machine learning—and a newfound momentum for relational databases even as NoSQL experiences its doldrums.
The report uses O’Reilly’s internal analytics from use of its Learning Platform services, including four key metrics: search questions, resource usage by topic, resource usage by title, and questions asked of O’Reilly Answers (which itself is an AI-powered content indexing tool spanning a range of media). Of course, the sample is limited to O’Reilly customers: “You could read this as a report on the biases of our customer base,” the report cautions, noting that that customer base includes “millions of developers, from well over 100 countries.” (The company notes that “there are certainly areas into which we’d like more visibility, such as the crucial Asia-Pacific software development community.”) O’Reilly also notes that “if resources don’t exist, our customers can’t use them.”
For this report, O’Reilly analyzed data from January through September of 2021 (and, for year-over-year comparisons, the same nine months in 2020).
A Trough for AI Interest?
In examining the momentum of AI, the O’Reilly report makes reference to the much-feared “trough of disillusionment”—a stage of the Gartner hype cycle that represents waning interest in the wake of failures and slowdowns in early applications of a new product or technology. “AI has certainly been hyped,” the report reads. “But is it heading into the so-called ‘trough of disillusionment?’”
“We’d say no,” it continues. “That’s not what our data shows.” The report acknowledges that use of content with “artificial intelligence” or “AI” in the title are both down (23% and 11%, respectively), but hand-waves those “relatively small and narrow” topics in favor of what it calls “the topic that clearly dominates this space”: machine learning, which is accessed four times more than the AI content.
And those title topics (“machine learning” and “ML”) are stable or growing, with usage of ML-titled content up by 35% year-over-year and more titles than ever addressing the topics. Specifically, neural network content usage is up by 13%, along with usage of content on reinforcement learning (+37%) and adversarial networks (+51%), but deep learning-related usage is down (-14%). “Interest has clearly shifted from general topics to specific ones,” O’Reilly says.
“We don’t expect another AI winter,” the report reads. “AI is too solidly entrenched in online business practices, and in ways that aren’t as visible as social media recommendations; you’ll never know (or care) whether the company that makes your espresso machine is using machine learning to optimize the manufacturing process and manage inventory, but if they aren’t now, they will be.”
The report hedges, though, that AI and ML themselves are “natural outgrowths” of terms like “big data” and “data science” that are now on the decline as terminology, despite, in reality, simply evolving into other terms and concepts. “The question for the coming year, then, is whether machine learning and artificial intelligence will “evolve”—and if so, into what?”
A Shift in the Winds for Databases?
“You can’t talk about machine learning without talking about data and databases,” the report reads. On that front, they say, “momentum has clearly shifted from the NoSQL movement back to relational databases.” This is a somewhat surprising claim, as lately it has appeared that NoSQL databases have been giving relational databases—the standby for decades—a run for their money.
First, the report hedges its characterization of NoSQL. “NoSQL was never a single technology,” it reads. “Databases like Cassandra, HBase, Redis, MongoDB, and many others are wildly different. NoSQL is really more a movement than a technology—one that’s devoted to expanding the number of storage options for system designers.”
Indeed, content on these NoSQL options varied wildly in trends: usage of content for MongoDB grew (+10%), but others declined: Cassandra (-27%), Redis (-8%), and HBase (-57%). All of this added up, O’Reilly says, to total usage 40% greater than MySQL, but which represents a 4% decline year-over-year.
Those faithful relational databases, however, are experiencing growth in content usage: Oracle (which the report says “is leading the pack” of database content) is experiencing slow growth consistent with its established nature, and usage of content for the (now Oracle-owned) open-source MySQL database tech has grown by 22%.
“Relational databases still dominate the database world, and there’s no reason to expect that to change,” the report concludes. “Nor should it. The promise of NoSQL wasn’t replacing relational databases; it was increasing the number of options available. The rise of graph and time series databases are simply examples of this promise in action. It will be interesting to see whether this trend continues into 2022.”
The report also notes that use of content with “graph databases” is up by 44% (“It’s a small category, but that’s a significant signal”) and “time series databases” is up 21%.
About the Report
O’Reilly’s “Technology Trends for 2022: What O’Reilly Learning Platform Usage Tells Us About Where the Industry is Headed” report was compiled by Mike Loukides. It is available to access at this link.