For a production server I think that the opposite is a better _general_ advice - reduce the available swap, because if your server gets to a point to need it, the performance will suffer so much that your server will become completely unresponsive. Having less swap will allow the OOM to kill the run-away process and allow you to login and fix the problem instead of rebooting the server or waiting in vain for it to recover by itself.

edit: typo