The collective strength of a system of fibers, each having a failure threshold drawn randomly from a distribution, indicates the maximum load carrying capacity of different disordered systems ranging from disordered solids, power-grid networks, to traffic in a parallel system of roads. In many of the cases where the redistribution of load following a local failure can be controlled, it is a natural requirement to find the most efficient redistribution scheme, i.e., following which system can carry the maximum load. We address the question here and find that the answer depends on the mode of loading. We analytically find the maximum strength and corresponding redistribution schemes for sudden and quasi static loading. The associated phase transition from partial to total failure by increasing the load has been studied. The universality class is found to be dependent on the redistribution mechanism.The response of an ensemble of elements having random failure thresholds plays a crucial role in the breakdown properties of heterogeneous systems [1]. Such systems can be, for example, disordered solids under stress, power-grid networks carrying currents, roads carrying car traffic, or redundant computer circuitry etc. The failure properties of these systems are strongly dependent upon the load redistribution mechanism following a local breakdown. While in some cases (e.g. solids under stress) such mechanisms are properties inherent to the system, in many other cases (e.g. power-grids [2], traffic controls [3]) it is a matter of design that can be optimized so as to achieve the most robust configuration (see e.g. [4]). Such robustness properties of networks connecting elements with varying failure thresholds under targeted or random attacks have received substantial attention [5,6].In this Letter, we address the question of maximizing the strength of a disordered system by finding the most effective redistribution scheme analytically using the fiber bundle model as a generic example. This model was introduced [7], indeed, to estimate the strength of cotton in textile engineering. Since then, it has been extensively studied in the context of distribution of failure strength and failure time of disordered materials under tensile loading or twist by viewing fibers as elements of the disordered solids having a finite failure threshold (or even a finite lifetime dependent of loading) [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] (see [23] for a review).Conventionally, the fiber bundle model is viewed as a set of parallel fibers, having failure thresholds randomly drawn from a distribution (say, uniform in [0 : 1]), clamped between two horizontal plates. When the bottom plate is loaded, some of the weak fibers break, and their load is redistributed among the surviving fibers, which may in turn break or survive depending on their failure thresholds. The load transfer rule, generally a function of distance from the broken fiber, plays a vital role (see e.g., [24,25]). While substantial attention has been concentra...