The brain uses mostly ketones to run itself on a ketogenic diet, but there are a few parts that need glucose. Usually these parts are the fate of the small amounts of carbohydrates consumed and the rest is synthesized from amino acids... or is it? Glycerol can be turned into glucose and there are a heck of a lot of glycerols in a ketogenic diet. How much protein actually gets lost? Do we preferentially convert proteins or glycerols when we need glucose?
Hold on, I think we got an answer here!
This experiment tested your question. After a normal meal, only 4.5% of total glucose production is from glycerol. After 3-4 days of fasting, that goes up to 21.6% of total glucose production. The authors go on to advocate the notion that rapid fat loss during fasting is primarily to provide you with glucose. I dunno about that, but 21.6% is a lot! That means maybe we can eat a little less protein on longer ketogenic diets, perhaps?
The second way to get protein outside of diet, as mentioned by Dr. Eades, is to get recycled protein junk from our cells. But how much protein junk could we have, and how long would it last?
Glycerol is ~10% of the weight of a long chain fatty acid triglyceride. So about 10g glycerol per 100g fat. I imagine that our metabolisms are "geared" to use something like "excess" glycerol to make glucose before using amino acids.
Using the numbers from Kamal's linked study, one could presume the "starvation" rates of almost 70% glycerol converted to glucose when in ketosis. By my math, that's still perhaps 10-15g glucose from dietary glycerol.
It's always been curious to me that our fat tissue cannot use the glycerol released when we break down stored trigs to re-assemble others. Perhaps it's designed that way to provide a certain amount of substrate?
what ever we release by lipolysis gets cycled as long as we have TCA intermediates.....if not we shunt them to pyruvate then to lactate. And for completeness our brains can use both as fuel sources. There is no limit as long as our mitochondria remain coupled.
My recollection is gluconeogenesis converts protein at 58%. Also, the amount of glucose from burning fat was a few percentage points, negligible for the most part. Unless you happen to be going through a lot of bodyfat in a short amount of time. The glycerols from fat are pretty much all the time from the bodyfat being used as fuel. The protein conversion is only for when the body needs it. Not so much a preferential thing per se as far as I understand it.