Dynamic spectrum management (DSM) has the potential to improve crosstalk-limited, rnulti-user digital subscriber lines (DSL) systems throughput. Unfortunately, an efficient, high-performance DSM technique has yet been discovered. Optimal spectrum balancing (OSB), currently the best available spectrum optimization technique, is capable of determining the optimal throughputs of discrete multi-tone (DMT) based DSL systems but it suffers with exponential computational complexity. The focus of this thesis is to identify more computationally efficient rnethods to improve the performance of OSB. First, a time-sharing condition, is observed for the spectrum optimization problem of OSB. This observation re-interprets OSB as a dual algorithm that converges to the global optimum of the spectrum optimization problem, and this observation leads to more efficient dual update methods. Next, this thesis proposes iterative spectrum balancing for approximating the dual objective more efficiently. Simulation results show that the proposed algorithm outperforms iterative water-filling (IWF) in different DSL scenarios.