The Early Germanic Dragon
Smaug of The Hobbit is arguably the most famous of all modern dragons, although his creation is heavily indebted to early Germanic tradition (which is hardly surprising, given Tolkien’s interests). Like his counterparts in Beowulf, Völsunga saga, and the Eddas, Smaug is a large serpent-like hateful creature with a vulnerable underside and makes his bed on a hoard of ill-gotten treasure, which he greedily guards. And like the Beowulf-dragon, Smaug flies and breathes fire.
However quintessential the dragon may seem to early Germanic legend and later spinoffs in the manner of Tolkien, it may come as a surprise to some that the usual word for ‘dragon’ in the early Germanic languages—Old English draca (masc. n-stem), Old Norse dreki (masc. n-stem), Old High German trahho (masc. n-stem)—is ultimately a loanword, from Latin dracō ‘dragon,’ itself a borrowing from Greek drákōn ‘dragon, huge serpent, python.’ (ModE drake is the reflex of OE draca, while dragon is a loanword from Medieval French, itself a reflex of Latin dracō.) In fact, the early Germanic languages lacked a unique inherited word for ‘dragon’ and—barring the derivatives of dracō—used a term that meant broadly ‘creeping creature, of either the reptile or larvae family’:
OE wyrm ‘reptile, serpent, snake, dragon, creeping insect, worm, maggot’
ON ormr ‘serpent, snake, dragon, worm, maggot’ (and the related yrmi ‘vermin’ < *gawurmija, a collective formation)
OHG wurm ‘snake, serpent, dragon, worm’
Gothic waurms extant three times with the meaning ‘snake, serpent’
The Proto-Germanic noun *wurmiz (with variant *wurmaz), whence all these forms stem and which must have also meant broadly ‘a creeping creature,’ evidently preserved an inherited meaning also found with cognates outside Germanic, at least in part, cf. Latin vermis ‘worm, maggot,’ Lithuanian varmas ‘insect, mosquito,’ Old Church Slavonic vьrmьje ‘insects.’
The Latin loanword never fully supplanted the reflexes of *wurmiz, but in OE and ON, the ‘worm’ forms when used to denote specifically a ‘dragon’ were largely restricted to a literary or poetic register. And this is still the case with the reflex in Modern English.
And why was the Latin word borrowed? The commonly accepted view is that dracō entered West Germanic initially as a technical term denoting a very specific kind of war-standard introduced into the Roman military in the second century AD by “Scythian” i.e. Sarmatian units (Homann 1986: 131; Lebedynsky 2001: 203-206). These ‘dragons’ (windsock standards with a dragon-head mount) are clearly described by Arrian (Tekhne Taktite 35.2-4) in the second century, for example, and are illustrated in a number of contemporary and later sources (see figure below):
In their charge the [cavalry] sections are distinguished by their standards, not merely Roman but also Scythian, so as to make the charge more colourful and terrifying. The Scythian standards take the form of serpents [drakontes] of even length and hanging from staves. They are made by sewing pieces of dyed cloth together, with their heads and whole body right down to their tails like snakes, so as to produce as terrifying a likeness as possible. When the horses are halted one sees these devices as no more than pieces of coloured cloth hanging down, but when the horses are in motion, they are filled out by the breeze and look remarkably like beasts and even hiss as the rapid movement sends the air through them. (Translation in Hyland 1993: 73)
Once the term dracō was adopted as a term describing a dragon-shaped standard, it evidently began to be used more broadly at the expense of ‘worm,’ eventually becoming the common word for ‘dragon’ in the Germanic languages.

‘Dragons’ from Trajan’s Column (Lebedynsky 2001: 204)

Reenactor with a “dragon.”
To see a reconstruction of the one of these “dragons” in action, with their howling sound, watch the section between 39:48 and 41:03 of this Time Team episode:
Bibliography
Homann, H. & T. Capelle. 1986. “Drache.” In Beck, H. et al. (eds.), Reallexikon der Germanischen Altertums. De Gruyter. 6: 131-137.
Hyland, A. 1993. Training the Roman Cavalry, From Arrian’s Ars Tactica. Sutton.
Lebedynsky, I. 2001. Armes et guerriers barbares au temps des grandes invasions. Errance.