Euclidean distance vs Pearson correlation vs cosine similarity?

Their goals are all the same: to find similar vectors. Which do you use in which situation? (any practical examples?)