Just ran across one of my favorite similes from Virgil, at the beginning of Book VIII of the Aeneid. But I should say something about similes in general first, because they are very strange things. This is especially true when they take the extended "epic" form, which I actually think is the purest. Now, there are a lot of reasons to think otherwise: the epic simile provides a comparison which does not so much compare as narrate another story in miniature, and when they are not annoying for taking us away from the action, they vex our attempts to make sense of them, as they force even best interpretations to turn towards the awkwardness of allegory.
But many of these reasons come from a sense of the rhetorical canon (it doesn't have to be even an explicit rhetoric) that subordinates the simile to metaphor. Never mind the fact that this tendency often comes from the modern sense of metaphor as a function rather than a--for lack of a better word (the traditional "trope" having been coopted by de Man et. al.) I'll just say an artifice--which tends to justify itself with references the psyche empirically understood. Whether these take place in studies that present the empirical psyche straightforwardly as such (in a marriage of psychology and poetics that begins with I.A. Richards), or studies that veil it in a pseudo-phenomenological garb of anxiety and trauma (in a deconstructivism or late-Lacanianism), such view is problematic--not because empiricism provides its foundation, or because the reference is too psychological, but because such it sees the place where the psyche meets rhetoric as language, which is thereby both kept obscure and granted too much power (it becomes Language). This, by the way, is why this sense of metaphor as a function is really only helpful in structuralist poetics, which interprets the psyche psychoanalytically (as in Barthes) or with a genuine phenomenological intent (Ricoeur often, or the more un-de-Manian aspects of Derrida). Situating the link in a linguistic system which, made up as it is of signs, is able to be studied (that is, neither undifferentiated or wrapped in the mystery of infinite differentiation), these stay true to the real aim of the functional sense of metaphor, best outlined by Jakobson: to recover rhetoric by recovering its explanatory power, which means making it more economical (two tropes, which are brought into closer relationship to schemes) so as to restore some sense of the urgency of debates over typology (it will matter again whether the instance in question is an instance of metaphor or not).
But never mind all that: the point is that everything that makes us subordinate simile to a metaphoric function doesn't help us when it comes time to actually get a sense of the purpose of the simile. Here, it is more helpful to turn things around and say as Pope once did that a metaphor is really just a little simile: this junks all the deeper things we have learned about metaphor, and reduces them to the comparative purpose of the simile, but it gives us a perspective that doesn't take this comparative purpose for granted. More significantly, it changes the relationship between metaphor and simile from one of explicitness (when we think of simile subordinated, we often say that it is just a more explicit metaphor: we are apt to explain it as a metaphor that just "has" like or as) to one of size: we thereby understand the comparison accomplished by the "littler" simile as something more like a short illustration, and the comparison of the actual simile as something elaboration. There is something disgusting to modern ears in this, because it comes close to the late-19th century finishing-school sense of such tropes as ornaments that make everyday speech more noble, decorated, and serve to puff it up. But comparison as elaboration is something different than comparison as ornamentation, and it gives us some sense of the essential role that earlier generations felt such tropes played. For they did not have such a disgusting sense that proper speech should be utterly unornamented, devoid of anything but the most rudimentary grammatical connections and the plainest, most dumbed-down meaning: they did not have the sense that meaning was utterly opposed to the means that expressed it (a sense which is only exacerbated by weak attempts to dissolve this opposition and make everything "linguistic," like those of the empirical/postmodern theories I mentioned above--which is why I complain about them).
If we view the simile then in this light, we understand more its relationship to narrative. Perhaps we even see it see it as a modification of the storytelling impulse itself, a modification which seeks to put the fictional aspect of stories to good use: it seeks to find what in fiction compares or elaborates reality, and uses it to work up a fiction already being elaborated. Such a role might actually bring it more in line with a different, lyric impulse, which does not oppose fiction and reality but sets them side by side: the simile might be lyric trying to tell a story, in other words, or narrative trying to return to its similar lyric-like--sorry for the similes--lack of opposition to reality (considered as something different from history, which is mimetically duplicated--more on this in a moment). This view of similes can perhaps most be opposed to view that has to make sense of them as allegory, relating the components together into a hard lump which, in its self-consistency, opposes itself holistically, at one go, to whatever is being compared: maybe it is even the trope that is the very antidote to allegory itself, being always closer (even in its more condensed, illustrative moments) to something like a parable.
Of course, what challenges such a view is the mimetic function that is most explicit in a simile. But what if we rethought mimesis in our rethinking of this very explicitness above, though we did so in a (supposedly--I'd say seemingly but that's what is precisely in question) different connection? Elaboration, like fiction itself, surely involves mimesis, but is not therefore an elaboration of anything that would bring it into opposition to reality, or, as was often said in the height of a celebration of postmodernity, undermine reality (for similar--sorry again--perspectives on check out the writings of Paul Fry and Michael Wood: Fry has been advancing this "realist" view since the 80's). Once we admit the fact that fiction, with mimesis at its center driving it on, is not as opposed to reality as it is to history (the weird internal timeframe of literature, which you can sit down and slip back into at any time, is an index of this--which does not, for all that, make literature itself something ahistorical), we begin to see how tropes that involve it, that mobilize it, might actually be the stuff of literature that is most in connection with the real.
This probably requires a bit more thought (how does literature's non-opposition to the real differ from its elements? how does a novelistic fiction's elaboration of reality differ from lyric's, which is obviously more direct? and how much is rhetoric fictional, especially when it is used in "non-fictional" [a pejorative term, that brings fiction back into opposition with reality qua history] discourses?). But in the meantime, it certainly shows you why I grant the epic simile primacy, as something like the "model" of all similes, and explains as well perhaps the most fascinating thing about epic similes (and perhaps all similes, then): their amazing translatability. Because from a practical perspective they serve to elaborate first and foremost (or tend to open up into something more than illustration), they gain a certain freedom from the selection of words that tends to modify illustrations (metaphors) more. They differ, of course, from translation to translation, but are wonderfully portable.
Perhaps I'll give you an example with the following simile, to which I finally come:
quae Laomedontius heros
cuncta uidens magno curarum fluctuat aestu,
atque animum nunc huc celerem nunc diuidit illuc
in partisque rapit uarias perque omnia uersat,
sicut aquae tremulum labris ubi lumen aenis
sole repercussum aut radiantis imagine lunae
omnia peruolitat late loca, iamque sub auras
erigitur summique ferit laquearia tecti.
Meanwhile the heir
of great Laomedon, who knew full well
the whole wide land astir, was vexed and tossed
in troubled seas of care. This way and that
his swift thoughts flew, and scanned with like dismay
each partial peril or the general storm.
Thus the vexed waters at a fountain's brim,
smitten by sunshine or the silver sphere
of a reflected moon, send forth a beam
of flickering light that leaps from wall to wall,
or, skyward lifted in ethereal flight,
glances along some rich-wrought, vaulted dome.
-Theodore Williams' (pretty literal) translation.
While Turnus and th' allies thus urge the war,
The Trojan, floating in a flood of care,
Beholds the tempest which his foes prepare.
This way and that he turns his anxious mind;
Thinks, and rejects the counsels he design'd;
Explores himself in vain, in ev'ry part,
And gives no rest to his distracted heart.
So, when the sun by day, or moon by night,
Strike on the polish'd brass their trembling light,
The glitt'ring species here and there divide,
And cast their dubious beams from side to side;
Now on the walls, now on the pavement play,
And to the ceiling flash the glaring day.
'T was night; and weary nature lull'd asleep
The birds of air, and fishes of the deep,
And beasts, and mortal men.
-Dryden's translation (I include three more lines because I think Dryden is doing a balancing act of some sort between night and day which nicely takes off from Virgil.)
I'll add other translations as I hunt them down.