>Unless you sync disk state to remote storage on every write, wouldn't you always lose some writes
You can trade durability for availability (the database isn't useable until the disk is available). You'd have some sort of redundant disk setup (on top of normal backups)
You run into the same problem with RDBMS like Postgres. If you enable synchronous replication, you go from 1 SPOF to 2 (both servers need to be available to ack a write or you lose your redundant data guarantee).
You can trade durability for availability (the database isn't useable until the disk is available). You'd have some sort of redundant disk setup (on top of normal backups)
You run into the same problem with RDBMS like Postgres. If you enable synchronous replication, you go from 1 SPOF to 2 (both servers need to be available to ack a write or you lose your redundant data guarantee).