[sgarland]

Databases. SQL isn’t going anywhere, and neither is Postgres. Plus the various NoSQL offerings.

DBA has been reinvented as DBRE. It’s essentially still the same job, except you also have to know how to wrestle YAML, HA/DR, and create meaningful SLIs / SLOs. But if you like getting your hands dirty with the internals of DBs, hoo boy are there jobs. My last interview run, I had a 50% callback rate (n=10), plus some extras where the recruiter hit me up on LinkedIn.

The job can and should entail Linux ops knowledge, so if you aren’t great at that, learn it to a great detail. Somewhat obscure things like filesystem and kernel tuning are helpful.

I say can and should because yes, if all you deal with are managed DBs, you can pretend Linux doesn’t exist – but it does, and it’s enormously helpful to have a good handle on concepts like IOPS (including how latency affects it, and how ops requests can be batched by the kernel [and EBS FWIW], etc.).

[sharadov]

Most companies have managed dbs, and they don't hire a DBRE/DBOps person until it's too late. But I agree there are jobs and not many people with the skills.

[sgarland]

> they don't hire a DBRE/DBOps person until it's too late

My pet theory for the explosion of jobs in this space is that during ZIRP, tons of companies underwent huge growth (generally without a DB team, as you said), and managed by just upsizing instance classes. Eventually, you can’t solve your problem by throwing money at a vendor.