Summary
In our production environment we saw some cases where applications will produce metric names with UTF-8 characters or with characters that are not allowed by original StatsD protocol specification. We could create some kind of strict flag which would drop samples if they don't conform to protocol.