We describe a study of optimization techniques which have been proposed in the literature for registration area based location management from the perspective of optimal on-line algorithms. We show that most of these optimization techniques such as forwarding pointers and overlapped registration areas can be modeled using metrical task systems. Some of these models are simple while others are somewhat intricate. These representations directly imply the applicability of existing results on metrical task systems to algorithm design for location management. However, this also means that general lower bounds known for online metrical task systems carry over to our formulations of location management. We also discuss some restricted models in which much better (and simpler) algorithms are possible.