When I say “reliability,” what comes to mind? In our industry, probably a couple of things: the reliability of your software, or the infrastructure it runs on. That makes sense- software has bugs and performance problems, and infrastructure gets complex and fragile when designed poorly or ignored as traffic scales. But here’s a question we don’t ask often enough: what about your DATA?

Your databases, your application state, your designs, your reports- the business doesn’t function without them. And most teams think they’ve got this covered. You’ve got replication. You’ve got some cloud provider offering you automated snapshots and eleven nines of durability. So your data is safe, right?

Not really. Replication will copy data corruption, ransomware, and your accidental UPDATE SQL statement without a WHERE clause. High availability saves you when a server or even a data centre dies- it does nothing when someone accidentally runs a stray rm -rf command. And the one thing that actually gets you out of trouble, restoring from a backup, is the thing almost nobody ever tests.

My guest today has been spending a lot of time staring at this problem. Gilles Chehade is an IT architect and senior lead developer, a long-time OpenBSD contributor, and the creator of OpenSMTPD. He’s also the co-founder of Plakar, an open source backup and restore platform.

Operating systems, email infrastructure, and backups- a man after my own heart.

Links: