News for hackers. Stuff that matters.
Re:AWS EC2/RDS Outage in us-east-1 by dekhn on Saturday August 31, 2019 @12:04PM (#20847108)
usually, DB servers will live in a small number of locations with good connectivity between clusters and the frontends (which terminate the user's TCP connection) live much closer (likely Sydney). Good design means that there are few roundtrips between the FEs and the backend but they are not unavoidable.

Designing truly resilient and available applications with DB servers that replicate across continents is hard.

Re:AWS EC2/RDS Outage in us-east-1 by juliusmusseau on Saturday August 31, 2019 @12:10PM (#20847159)
Is true master-master replication across continents even possible?

I guess partitioning can help, but then isn't it just turning the DB servers into pizzas of master-slave where the Hawaiian slice is master only in Hawaii, and slave everywhere else?

Re:AWS EC2/RDS Outage in us-east-1 by yjftsjthsd-h on Saturday August 31, 2019 @12:15PM (#20847195)
Yeah, you're gonna hit CAP hard at that distance.