Ronald Marsh* states that after 175 there was systematic fortification of the town. What is known of the walls indicates that the castle hill was within them. Brown** states that the castle used the remains of the Roman town walls as foundations. There is no evidence of a formal fortress (though strong supposition of Plautius setting up a small fort during the invasion) but the fortified town enclosed the castle hill. Subsequent Norman building has erased the previous property there (from Domesday we know that the land yielded an annual rent of 17/4 to the bishop in 1086 values). That's why I used the term "strong point" and eschewed "stonghold", "fort", "fortress" or "castle".
Pre-Roman Belgic remains and coin dies have been found. The Celtic "kingdom" of the Cantiaci had two main administrative centres or 'oppida': Canterbury and Rochester. Due to two millennia of changing land use there is no evidence whether this was a true oppidum (a large fortified Iron Age settlement) or something much smaller. Hence my final warning sentence about "How much importance and how much size is, however, pure speculation and nearly devoid of evidence!"
There is a suggestion that the Watling Street is based upon an earlier trackway which crossed the Medway at this point. I've not been able to find any details of how, there wouldn't have been a bridge and I would have thought the bottom too silty for a ford. However, at that time the currents and depth were less (Strood hadn't been built up narrowing the river) so a low tide crossing might have been possible, alternatively there might have been some sort of ferry. The river used to cut in to below the castle (until the Esplanade was built) and early drawings show it being used by boats as a landing place.
*Marsh, Ronald (1974), Rochester, The evolution of the City, Medway Borough Council
**Brown, Reginald Allen (1969), Rochester Castle, London: Her Majesty's Stationery Office