Cassandra and The trouble with timestamps

http://aphyr.com/posts/299-the-trouble-with-timestamps

"This behavior violates strong, eventual, causal, read-your-writes, session, and monotonic write consistency, and depending on how you interpret "seen", violates monotonic read consistency as well."

Thank you kind author!

I think Cassandra is a really neat piece of software, and it's certainly come a long ways. But this Timestamp issue has always given me pause because not only can I get completely broken results, I also can't even audit when it happened, is happening or will happen.

I'm sure this has happened: 

"We have a redundant array of nodes, storing 3 copies of data across the cluster. Each node runs RAID 10 ensuring high performance and provides excellent data protection, requiring a minimum of 6 drive failures before losing data!" 

"Unfortunately last week the was a single hardware failure in the system clock on a single node, so all our data is garbage."

OTOH, I can imagine saying something similar for a lot of software I've written if the RTC is truly broken, and Im sure Oracle has similar issues. Still, Cassandra timestamp failures seem to be in a different class.

-JD