Good point.  When I think of scaling, I think in terms of things like ease of clustering/load-balancing, use of centralized/shared caching (eg memcached), robust DBI connections for different configurations (eg Master-Master or Master-RW with a cluster of Slave-RO nodes), etc.  Since Dancer is so good at remaining a micro-framework, many of these choices are in the hands of the coder rather than these choices being made for them by the framework.  Since Dancer doesn't dictate those choices (thank you!) maybe the Cookbook or some Plugins could fill in the leap of imagination required for those who want to envision Dancer in a scalable context.   By providing some best practices, cookbook recipes and deployment howtos, the Dancer community gets to define what scaling means to Dancer.

My two cents.