Despite the Romans and the Roman army having numerous dramatic large-scale battles in which thousands upon thousands died (Cannae, Allia, Tuetoburg) many Roman soldiers survived long into veterancy and even into retirement.

This is especially the case during the Pax Romana, in which, as you've alluded to in your question, soldiering was a career, rather than a civic duty. This allowed soldiers to increase their skills / effectiveness over long periods of time, which saw them through a lot of battle.

To quote Erdkamp's "A Companion to the Roman Army" (2007):

"a simple model life table for adult males suggests that of 100 soldiers who enlisted at age 20, 78 would have survived to age 35, 69 to age 40, and 60 to age 45. For legionaries, this implies a baseline rate of attrition of roughly one-third for 20 to 25 years of service. In reality, violent death, camp-related disease, and early discharge would have raised overall attrition by a potentially significant margin."

Alston, however, writes in "Soldier and Society in Roman Egypt" (2002) that legions serving in Africa discharged "between 67 and 390 veterans" (p. 43) every year. He continues to add that by looking at the number of recruits added to legions each year you could approximate the losses in a legion (this is of course not very scientifically accurate but it gives you an idea). The figures he found were 250 new recruits and 150 veterans discharged per year.

That gives you a number of 100 (on average) dead (p. 42) - though keep in mind this might not be through battle.

Out of a legion size of 5,500 (for the first century AD, anyway) the number is quite a lot less than I suspect most would imagine.