Methods and apparatus for cost minimization of multi-tiered infrastructure with end-to-end delay guarantees

Apparatus for designing capacity allocation for a multi-tiered computing system, each tier of the computing system having one or more computing devices, the apparatus comprising:

• a memory;
• and
• at least one processor coupled to the memory and operative to:
• (i) obtain input parameters relating to capacity allocation;
• and (ii) compute a capacity allocation based on at least a portion of the input parameters utilizing a nonlinear program, the capacity allocation being computable such that one or more mean delay measures and one or more tail distribution measures are satisfied for multiple service classes, wherein the step of computing the capacity allocation further comprises formulating a capacity allocation problem as an optimization problem so as to satisfy the one or more mean delay measures and the one or more tail distribution measures, and wherein the optimization problem is formulated as a separable convex programming problem with linear constraints which is solvable by finding an optimal solution on a boundary of a convex hull representing feasible solutions to the programming problem in accordance with a location of the convex hull in relation to a feasible hyperplane, such that the optimal solution represents the capacity allocation that is implemented for the computing devices of the multi-tiered computing system.

An article of manufacture for designing capacity allocation for a multi-tiered computing system, each tier of the computing system having one or more computing devices, the article of manufacture comprising a computer readable storage medium containing one or more programs which when executed by a computer system perform the steps of:

• obtaining input parameters relating to capacity allocation;
• and
• computing a capacity allocation based on at least a portion of the input parameters utilizing a nonlinear program, the capacity allocation being computable such that one or more mean delay measures and one or more tail distribution measures are satisfied for multiple service classes, wherein the step of computing the capacity allocation further comprises formulating a capacity allocation problem as an optimization problem so as to satisfy the one or more mean delay measures and the one or more tail distribution measures, and wherein the optimization problem is formulated as a separable convex programming problem with linear constraints which is solvable by finding an optimal solution on a boundary of a convex hull representing feasible solutions to the programming problem in accordance with a location of the convex hull in relation to a feasible hyperplane, such that the optimal solution represents the capacity allocation that is implemented for the computing devices of the multi-tiered computing system.