[DaiPlusPlus]

Yes, but consider the marketing, uh… I mean, messaging coming from the NoSQL and DBaaS (CosmosDb, DynamoDb, etc) and Firebase crowd: “databases are hard, let us manage it all for you” - and they’ve got a point: it’s 2024 now, we arguably shouldn’t need to handle those kinds of non-functional requirements by ourselves: a DB engine should be able to automatically infer the necessary indexes from the schema design, and automatically rebuild them asynchronously - if Google can search the web in under a second, then your RDBMS should have no problem querying your data in a fraction of the time.

…which I imagine is the impression made to a lot of (even most?) people who got started writing software-that-uses-a-database within the past decade. If you’re using NodeJS then using a KV-store library feels far more natural than writing SQL in a string. At least the kids are using parameters now, so it’s not like how every PHP+MySQL site was vulnerable to injection attacks…

(I know that recently RDBMS now do implement automatic indexes based on runtime query-profiling, which is great, but it isn’t pre-emptive, and often gets it wrong too)

———-

Also, who calls themselves a “DBA” anymore? That word makes me think of a pipe-and-suspenders type, still employed well-past retirement age because they’re the only ones who knows how to keep the company’s Big Iron (…or AS/400) database from keeling over. Thesedays it’s all “Ops” - “DevOps”, “SysOps”, …”DatabaseOps”?

[teaearlgraycold]

Google doesn’t search the internet in under a second. They are doing a key-value lookup. Any time they change the search algorithm or add to their index they recompute every query ever searched and prepare the answers to be quickly recalled.