Reverse-engineering Roman cavalry
structuralarchaeology.blogspot.com
structuralarchaeology.blogspot.com
Space inside the walls would be at a super premium, so they would also likely need turnout paddocks for peacetime exercise, breeding, and nursing, which implies the fort would be surrounded by some fencing or corrals. They need lots of water as well, so there would have to be a stream or aqueduct within convenient human walking distance. The (huge) muckpile would need to be far from the stream and food, but accessible as fertilizer for crops. Maybe people outide the walls collected it, also implying roads and cropland nearby.
Sentinels every half mile or so would be able to sound an alarm to get everyone inside the walls in case of an attack, and you can pretty reliably call in a whole herd to run back to the stalls with a signal. Even faster with working dogs.
The standard exercise arena size today (2500 years later) is 20x60m, which is about 65x200' and cavalry would need to exercise each horse in something almost that large, especially given the implied stall size and limited options for turning out for grazing.
Only reason I've thought about this is I am familiar with Xenophon's work, not just On Horsemanship, but also the Cavalry General as a rider: https://www.gutenberg.org/files/1172/1172-h/1172-h.htm
Super interesting.