convolution versus max pooling

my understanding is that convolution kernel is a sliding window method, but max-pooling is a deserted down-samples method.