While some caution is good, more isn't always better.
I have typically allowed limited communications from external hosts and devices in to management systems via well defined and understood paths. You can certainly (and probably should) NAT the target and apply the necessary ACLs (with logging that you or your organization actually checks regularly) to ensure that only permitted communications are allowed through your controlled interface.
Hope this helps.