@PaulWhite, I thought you made a compelling argument about the spaces vs tabs. So I wrote a crude script to replace every tab in the code fence with 4 spaces (and remove trailing whitespace as well)
Cassandra was an oracle in anciety, daughter of Priamus, king of Troy. Seemingly good name for a database, but she was famous for predicticting disasters.
and now China has announced the PLA will conduct exercises in six areas close to the island of Taiwan from August 4 to 7 with live fire drills. What could go wrong.
It also gets so crowded - Pisa was crowded too. But I very much enjoyed Siena and San Gimignano and Certaldo. And my long walks around Petronagno and Val D'Elsa
"Obsolete Assumption: Databases Need Reliable Storage" yeah floppy disks here we come. The fact that you have EBS or whatever other cloud storage doesn't mean you don't need to commit the data to some sort of storage. If you can't guarantee durability then you don't have data to store in the first place.
"Obsolete Assumption: Your Storage Is Slower Than Your Network" hmm anyone running 40gig in their servers? That's passé these days. Even CPUs struggle to keep up. And you can still cache, you just have to be aware of cache invalidation, as always.
"Obsolete Assumption: RAM Is Scarce" depends what size data you are talking. Terabyte databases are not big these days, and you'd struggle to get many servers that can manage that. "AWS gives you tremendous amounts of RAM for a pittance." Oh so are you paying my bill? They make a killing partly because to get more RAM you need to up the CPU massively.
"relational databases originally assumed they had reliable physical disks attached" no they didn't. HDDs were far more unreliable than modern SSDs or even modern HDDs. The whole point was ensuring you could back it up and get a consistent backup.
"for instance by requiring a caching layer to deliver the speed that could be obtained cheaper and easier with fast local storage" erm local storage is a caching layer. Scale-out is just putting the same data multiple times in multiple places, and every good RDBMS has done a lot of work on that.