-
S3 bucket and filename for config should be specified as part of the CFT and learned via a runtime parameter of some sort.
-
Better error handling
-
Host-down confirmation from within the Lambda service. Essentially, the idea is to treat host-down messages as prospective HA events, confirm them via the Lambda service, and only then initiate a route-table failover.
-
Configuration UI of some sort