I did Milano-Siracusa not many years back, following on a few hours after arriving from a London-Paris-Milan journey. Sleeper was fine; morning espresso excellent. I didn't mind the lack of restaurant/bistro/whatever carriage in the evening, having had a good meal in Milan in between trains, and it wasn't too early in the evening setting off from Milan. But I'd have appreciated (qualitatively and variety-wise, if not quantity-wise) having food beyond my own supplies for the morning run down the southern coast to the ferry. (I'm pretty sure there was a shower as well as a loo at the end of the carriage - though I might be mixing it up with some other overnight journey.) When I did this, the cost including a berth in a 2-berth sleeper was barely more than £100 per person from all the way from London to Syracuse, booking months ahead. But Eurostar is more expensive now; as might be Paris-Milan and the sleeper.
The whole 30+-hour experience, door to door, was overall pleasant and civilised, though - as mentioned - assume the Italian sleeper might not arrive quite on schedule! Given the lack of other train facilities, then that 1000 miles Milan-Syracuse (c 20 hours I think it was) is about the limit of what I'd want to do as a single leg of a journey! However, being able to get out onto the boat briefly, while the train was being ferried across the strait, just about broke up the last part of the journey sufficiently.
There's also the option - if you're in Milan already - of doing the journey time-shifted forward some hours, seeing some of northern Italy in daylight but sleeping through some of the south past Naples. The problem with that is that the Sicily sleepers don't start as far back as Turin/Milan so it requires a change. You either have to get an ordinary train down to Rome, and pick up a sleeper there in the evening; or get an afternoon sleeper in Milan but change for a train across to Sicily when you get to Villa S Giovanni first thing in the morning (the afternoon Milan sleeper continues south on the mainland).