Autoscaling Sync Gateway on AWS

My app, like many, has a wide variance in usage throughout the day. I’ve created a functional load balanced Sync Gateway pool on AWS with health checking in place. If server is unreachable, a new instance with custom AMI is brought online to replace it. So far it is working very well in beta. While I can manually increment and decrement the number of servers in this pool, I’d like to automate the process before adding a large number of users.

What would be the best (preferably single) metric to monitor to determine when it would be best to add an additional Sync Gateway? CPU usage, latency, etc? Combination of measures?

Sounds like a cool project! I bounced this off some of the other Sync Gateway devs, and the consensus is that CPU usage is probably the best single metric to look at.