Maternal inheritance of mitochondrial DNA is very well conserved, although some species, such as some mussels, show paternal inheritance. As for why or what the advantage is, some of it is due to basic logistics: sperm cells have ~100-1000 mitochondria, egg cells have 105-106, so male contributions are largely washed out. Plus, most mitochondria in sperm are toward the tail, which does not always or necessarily get inside the egg.
For a more selection-based mechanism of the actual destruction, think about what sperm do. They are small packets of energy that do nothing but race until they die. That's it. Energy production, which occurs in the mitochondria, produces reactive oxygen that can damage genomes. The mt-genome in the egg is far less likely to be damaged.
Additionally, I might look to chromosome uniformity. Heteroplasmy does occur now and again, and it's not usually a good thing. Mitochondria are very important, so it makes sense to have an all-or-nothing approach, lest any deleterious genomes persist.