I have never seen a system where the ControlLogix traffic actually overwhelmed the network capacity, and we can postulate that the network and devices on the network will work steadily when network conditions are stable.
I start with a raft of questions:
What is the originating device ? (1756-ENBT, -ENET, 1788-ENBT, SoftLogix 5800 ?)
What are the target devices ? (1756-ENBT, 1794-AENT, PV Std ?)
How many packets/second are being served by the originating device ?
How often does this problem occur ? Have you written "trap" logic to find out ?
How many unscheduled devices are also using the network ?
Any evidence of noise in the originator or target port statistics ?