There's a lot of recent discussion I'd love to weigh in on with an essay, but I don't know that it's germane to sizing the Excelsior, so I'll restrain myself unless asked by the thread starter. *lol* But as to the longevity of the class, I have some thoughts...
Starfleet seems to dump a lot of time and resources into rolling out some new linchpin flagship class every forty years or so. The development process yields ship classes from about halfway through that period that start having features of that eventual cornerstone Starship -- such as the New Orleans and Nebula having all those Galaxy-styled components before the Galaxy class itself was laid down (yes, I stick with registries being chronological to when a ship was ordered).
Kirk was head of Starfleet Operations for two and a half years prior to TMP -- which itself was only about a decade before the TWOK/TSFS/TVH block, when the Excelsior was launched. He would have known about that development project and the technologies associated with it. The class was being developed in large part because of simmering hostilities with the Klingons. Starfleet needed a new and badass Heavy Cruiser to replace the ageing Constitution class -- which, while still perfectly serviceable, was no longer a bleeding-edge front-line starship.
It makes sense to me that while various Constitution/Enterprise generation designs might still be in service for decades (Miranda, Sydney, etc.), their roles don't demand as much peak performance as the progenitor class. So when TUC happened, I can see Starfleet mothballing the Constitutions as a gesture to the Klingons, whilst keeping the new Excelsiors that they'd only just got off the skids. But since they were no longer needed as battleships to fight a potential hot war with the Klingons, Kirk -- after his replica Enterprise was retired with the rest of its class, and due to his familiarity with the Great Experiment from before -- led the team that helped turn the class into something new... Starfleet's first Explorer.
I... have many thought about the class, its design, its features, its capabilities, its variants... I also treat stuff seen onscreen as "official-ish". I know the shortcuts taken, last minute budgetary requirements (to this day, I insist the Oberth class is not the class the Grissom belonged to, just because Paramount didn't want to spring for the requested new ship class for "The Naked Now"). I have always loved the Excelsior, but also felt she was "unfinished". There are so many elements that make no sense for a fully completed starship design. That fantail is unnecessarily long and flimsy for so little apparent use. All the exposed ribs, and the cut-away tops of the nacelles seem like hull plating left off for access to structural elements. The whole warp engine assembly seems rigged for a quick jettison in case of runaway. I like to think a "finished" design plated over the tops of the engines, the neck, had pylons faired directly into the secondary hull, covered the open space between the aft shuttlebay and the neck, above, and extended the underbelly curve, below. The ship Sulu got was a hot rod. A bit like how CV(N)-65 was unlike any of the nuclear carriers that followed her.
Sidebar: The reasons so many consider Transwarp drive a failure is because it's never mentioned after TSFS until -- *cough* -- "Threshold". Meanwhile, in TNG's "Evolution", when the nanites are messing up the computer, Data mentions the last time there was a full system failure aboard a starship was over seventy-nine years previously. Which is right around the time of TSFS. Everyone also conveniently forgets that Scotty sabotaged the Excelsior's drive computer, and that it didn't just spontaneously not work due to bad design or bad science. Since the latter TOS films used the TNG warp scale, starting with TVH, some of us just think Transwarp succeeded, necessitated a recalibration of how we understood warp velocities to work, and just became the new normal. Until the term got used again in TNG and Voyager for very different things, indeed. I tend to think it's just a catchall term for something technologically within the "warp drive" paradigm (i.e., not quantum slipstream or other exotic things), but substantially beyond what was currently/previously technologically possible.
So this sturdy class was designed and built to be able to go toe-to-toe with Klingon battlecruisers and survive -- and then not needed in that role. They are, to me, the B-52s of Starfleet. Many classes have come out since, several flagship classes have introduced new technology and new techniques -- all probably intended to supersede the Excelsior class. But it just keeps being obstinately durable. I love the idea of something in Starfleet like what has happened in the real world: Air Force crew members flying the same B-52s over Afghanistan that their grandfathers flew over Vietnam. The Hathaway was mothballed in place in a remote star system after having been launched sometime circa the 2290s. The Hood and the Repulse are still pounding away doing all the necessary but unglamorous work of Starfleet, despite Captain Do Soto's tongue-in-cheek grumbling, and they're of similar vintage (yes, I go with the registry on the miniature for "Encounter at Farpoint" over the later Okudagram registry).
How long a starship lasts definitely depends in large part on what its intended role is. The Lantree started out in the 2260s as a state-of-the-art science vessel. A hundred years later, it's a sparsely-crewed, third-string supply ship. There's a sort of longevity in that, but not a lot of glory. The more a class has to rely on being top-of-the-line, the less likely it will be to stand the test of time. Similarly, classes rolled out for wartime will face a coin toss as to whether a use is found for them in peacetime. Smaller classes will likely stick around longer than bigger classes, as the resource demand of building a large-but-outmoded starship is harder to justify than a smaller one. Hence, new-built Mirandas well into the 30000s. So, to me, the Excelsior class is a bit of a fluke in that it was so well designed (or over-designed) that it was able to keep getting uprated to stay relevant for it to warrant Starfleet building more of them for decades, even after the next big flagship class -- the Ambassador -- came along in the 2320s.
(Oh, and as for size, I feel -- for many reasons -- that scaling it up about the same 15% as the TOS and TMP Enterprises need to be scaled up from their designers' given sizes works the best. That would put it around 535-ish meters -- but a lot of that is those loooooooooong engines.)