From: John Cowan <xxxxxx@ccil.org>
Date: Wednesday, July 10, 2019 5:50 PM

Or just forget disks altogether, keeping everything in the memory of several servers and just backing up to disk periodically.  See <http://delivery.acm.org/10.1145/1060000/1059805/p30-gray.pdf> (2005) for various arguments that disk-based row-oriented databases are no longer the be-all and end-all.

There's a place for all of the above, e.g. SQLite is heavily used client side, in Android, Firefox and respins of it, various minor client programs I use like xmms2, etc.  I can't get your URL to load, but with Gray almost certainly being Jim Gray, it's worth considering, I've read that Netflix uses this exact approach because AWS' other persistence methods were and probably still are awful or not appropriate like S3, using at least 3 servers each in their own Availability Zone (i.e. same region, different flood plains.

The Scheme community, though, doesn't exactly have those sorts of resources, we seem to be hard pressed to keep more than one machine/server running at a time.

- Harold