It’s doable, but it is hard, for a wide variety of reasons. To illustrate a few of the things that makes developing an MMO tricky, let’s analyse it backwards:
In a running game, you’ll have a large community. They each pay e.g. $10 pr month for the privilege (Which is why MMOs are compelling). This income is what you use to pay for maintaining the service. You use it to pay for bandwidth, server rent or depreciation, server maintenance staff, payment provider revenue share, GM staff, community moderators, website maintenance and a number of other minor stuff.
Assuming you want your game to live long, you want to fix bugs, improve performance, fine tune hardware usage, and generally fuss around the games infrastructure to optimise. This means a non-trivial amount of post launch purely technical development.
You’ll likely want to extend and adjust the gameplay and gameworld as well, adding game systems, nerf objects, add objects, add quests add troop types and whatnot. This is basically an extension of the game development procedure you undertook while developing the game, although at this stage, it’s a bit more tricky, because if you introduce a crashbug or an stupid quest or an overpowered sword of goblin slaying, and introduce this to your customerbase, they will be pissed of, and some will leave, costing you money. When you correct the mistake, some will feel nerfed, be pissed of, and leave. How many, depends very much on how well you customer relation people, GM staff and PR staff do their work.
Of course, you have to set the above up to work 24/7 365, no exceptions. This typically means staff in three shifts, which is mostly practical by opening branches in the US, Europe and Asia somewhere. (this way, you avoid having a “night shift”: Asia covers for Europe etc).
Luckily, you had some training, because all of the above, except payment, you need to have in full operation during your beta testing. that six month period, where people play your game without paying for it. that six month period that doesn’t end before you can measure on your player behaviour that they love the game, and would gladly pay the monthly fee to keep playing. Of course, marketing and PR need to get to work a long time before the beta starts, so you can get enough people interested in spending 10+ hours pr week on a promising, but buggy as hell game. Probably, PR and Marketing need at least a 3 month headstart.
Thats nine months, and we haven’t even touched actual game development yet.
Probably many will by now say “Yeah, but this is how WOW and WAR does it, we’re talking about small-scale MMOs.”
Well, first of all, WOW and WAR are on a different scale altogether. Their operation makes the above seem like a news stand in comparison.
Can you avoid some of the above described work ? Probably. Can you cut the Beta shorter ? certainly. Can you make do without PR and gamemasters ? sure.
Will you succeed ? probably not.
Does it make sense to make a small scale mmo ?
Absolutely.
If you can make a server setup that works robustly with say 3-400 simultaneous players, you can have a community of about 3500 subs. that’s a healthy amount of players, giving you a nice steady income.
Is it enough to pay for the staff, hardware and bandwidth ? good question. Bandwidth shouldn’t be the problem, nor hardware. Salaries might be.
Assuming the 3500 subs, you’ll have $ 35000 pr month. thats about three people, plus your hardware and bandwidth covered, in broad numbers.
Three people to operate an MMO is not a lot, not mentioning expanding the game, and caring for the community. But yes. it’s doable.
at least… it’s been done 
