Keith Darden points out that if Crimea secedes from Ukraine, electoral outcomes in Ukraine would shift with fewer pro-Russia voters in the political system, and that would be bad for Russia. This is not unique to this case.
In any successful secession or irredentist effort (the latter refers to annexing a "lost" territory inhabited by ethnic kin), the boundary moves, changing who votes in the rump state (Ukraine in this case) and who votes in the new state (Crimea if independent, Russia if this is irredentist and Crimea gets annexed). In this case, new voters in Russia would be largely irrelevant, given the populations of Crimea compared to Russia. But in Ukraine, the change is significant, given that the elections in Ukraine seem to have a blue/red, polarized, narrow outcomes kind of thing that might seem familiar to the US. Indeed, if Texas left the union (please?), and the electoral votes of Texas went away, the Republicans would have a much, much harder time winning national elections.
This change in the balance of political power is what can make one secession so very relevant for the rest of the country (see Slovenia and Yugoslavia). It also can be a deterrent to irredentism since the new voters may not vote the way the folks in power would want them--hence no Greater Albania project (a Greater Kosovo project is something else).
My point here is that Darden's point is not just relevant for Russian interests today but is relevant for understanding secession and irredentism in general--you are not just changing a line, and not just changing who governs person x or group y, but also who wins and loses elections (or other ways to allot power) in the new and old states. And folks anticipate that, shaping their preferences towards secession and irredentism.
Of course, that is not the only set of interests in play or else Harper's Conservative government would want Quebec to secede since that would mean a heap of Liberal and NDP voters no longer counting.... Hmmm.