The main departure point is that slavery in post-imperial Romania tended quickly to be mixed up with other servicable or client categories : Merovingian
pueri were generally lumped with
servi without much second tought, and
servi went from a metaphorical use for servents to a generic name.
At least North of Alps, the social situation was vague enough to avoid "warring slaves" to really blossom, especially as contrary to the Arabo-Islamic world (and somewhat ERE), service to the king or the head of a family was considered the privilege of the free man, what we use to call the vassalic link.
Now, it could happen in Mediterranean and Western Europe, sort of.
@Yanez de Gomera is right to point that Siculo-Normans did used a subservient Islamic archery for decades and while it's not really what you could call a servile army, the concept of subservient army, directly tied to the royal figure as "protected household" is interesting.
Visigoths, at least until the VIth century, kept a large part of armed slaves or semi-servile clients. T
he Visigothic Code pretty much clearly required any large slave-holder to arm and bring with him 1/20 of his slaves in battle. It seems that at least Burgundians did pulled something similar.
The aformentioned pueri, while not servile in the sense we could give to the word, were also a common part of Merovingian armies in the same period : as the name implies, they were originally non-veteran troops but young enough to be servicable. It of course changed a lot with time.
Let's remember that the institution of service created by Romans to refugees or deportees within their territory was carried on (or expanded) by Franks : Armorican Saxons, Continental Saxons, Alans, Alamans, Thuringians and other laeti/foederati were part of the Merovingian armies. But again, we're on a tradition of domanial/state service trough personal links, and it's the ground we have to use to create something akin, but vastly different from the "slaves on horses" of the Arabo-Persian tradition.
There I think
@Yanez de Gomera is wrong to tie the change with feudalism as it appeared in the late VIIth century (unless he meant the mix of manorialism, vassalic ties, Barbarized imperial institutions, etc. but it's quite confusing and not really relevant, as I tried to point above). It's rather how to make the Late Antiquity uses lasting ITTL, and I think it means at least no-Islam PoD and no rupture of the post-Imperial world, and possibly a Gothic hegemony in most of the region (less for political ties, than to highlight the 1/20th rule).
It's pretty much a given, tough, that the European model of subservient armies will be less of a household army, than a domanial army, and tending to mix up with the vast ensemble of serviciable or client classes eventually : still, it could make the development of a
miles sub-class interesting, as even more tied to lower classes than IOTL.