Over “millions of years” it does sound a lot doesn't it! When Darwin's The Origin of Species was published perhaps it seemed ample time, for some factor to be selected over millions of years.
Say we have the favourable mutation in a single individual in a population:-
Note, for evolution the population resets to one at each mutation, time would be required for numbers to increase.
Note, that while numbers are very low survival is at risk from minor local events, a local fire for instance. Risks become more normal when a population has spread out over a wider area. The individual faces death/elimination by normal causes which probably have nothing to do with the mutation, worst cases are for prey species. Every new mutation runs the risks of low population, this means every specie at risk today because of low numbers, would have been in that situation at least once before!
Whatever the amount of offspring required to produce a pair for the next generation breeding.
For example; 100 offspring produce only two for next breeding, then the mutation being on one sex or the other has only a one in 50 chance of survival (50 male or 50 female). So the mutation needs to occur (on average) 25 times for one to produce the next generation (for one in 50, could be any number between one or 50, Ave 25). The mutation may improve these odds or not, however it demonstrates the logic; many beneficial mutations would be lost.
Population increase may be slow due to close competition from the original form, the same in all but the one detail, competing and in numbers filling the available specie echo niche.
Just assuming each beneficial one survives grossly underestimates natural risk.
Then the favourable gene needs to be on a chromosome without another unfavourable gene on it, requiring elimination. There are several hundreds or thousands of genes on a chromosome and often several chromosomes in the make up of an individual, this favoured gene will be bundled with many thousands of others together. The whole bundle survives or not on the fortunes of the individual life form. This cannot work with experimental cleanness, if the situation is confused no effective result could be obtained, therefore if it ever works it must be with all non harmful plus one beneficial. Natural selection would need to be sequential, one at a time and this would maximise the timeline.
After the success of a single implementation, the new form must increase in numbers to occupy the specie eco niche, if it does not increase significantly there will not be the chances for another mutation. If the chances required for a single mutation without a bad, are ten million to 1, then the population needs to increase to that sort of scale. One wonders how small populations or recessive genes ever manage?
[additional: The chances are mutations are many more times likely to be harmful than beneficial, therefore this means in practice large portions of time to eradicate lodged harmful and useless clutter from the genetic system. If a mutation were a common event then a good mutation on a clean genome would be rarer, because if two are on the same package chances are one is bad, therefore the event is wasted. The imbalance that causes a mutation in the first place must fortuitously cause just the one without a harmful on the same DNA packet. Just one gets less likely as the gene quantity increases. In how many cases will the size of the life forms gene pool be greater than the whole population of a generation (genes/individuals)? Therefore evolution would slow as the genome increased in gene numbers.
[additional: I am not entirely content as yet with the notion of DNA chromosomes being the final secret of life, suspecting that the life functions in control during meiosis nearer that secret, the chromosome DNA may be for the adaptive and backup data structure. These could turn out to be same/similar/separate so wait and see, there is another box to open yet. However it does give us a way to quantify and scale.]
[additional: Similar to the above, I do not have the expertise to do other than suspect; is the virus a life form? Believers and Evolutionists are predisposed to different answers (' I perceive the elephant is like a tree' Indian story) . Evolutional thinking would look for the smallest life form to solve the sourcing mysteries, believers may find the smallest life form of interest as to, what is its place in the arrangement?
Because of the cellular attack nature of the virus; 1/ it is dependent on higher life forms existing, 2/ so far, it seems to give little evidence that, an independent life form can be that small. 3/ Is this nature of the virus similar to other functions in Nature?
Could it be the virus is a particle issued by a larger life form, the ultimate antibody, an immune system smart bomb? The changing nature of virus forms would then not be mutation but another issue by another individual life form. Perhaps the plan being the virus works locally; individual or ' touch company (a brood)' and then after killing off the target bacteria/parasites/intruders/tumour, the virus expires as no longer able to reproduce?
Just following my own logic. If created can we suggest why? I do not know the answer as fact but can see a potential why.
And maybe we should consider the logic for the blind men in the Indian proverb: All start asking All the questions. Does this make sense created? Why? Does this make sense evolved? Why? Is this a single item or part of a larger whole?]