In this article, the optimal distributed state-feedback control of interconnected systems is addressed. The underlying problem is nonconvex in general and does not scale well with system dimensions. The objective is to propose a scalable algorithm for distributed $\mathcal {H}_{2}$ and $\mathcal {H}_{\infty }$ design with the arbitrary control structure and limited model information exchanged within neighboring subsystems. To this end, first, a graph-theoretic perspective is provided for the structured optimal $\mathcal {H}_{2}$ and $\mathcal {H}_{\infty }$ design with linear matrix inequality constraints. Chordal graph decomposition is then utilized to decompose the underlying design problems into a set of lower dimensional problems with coupling constraints. Finally, scalable algorithms based on the alternating direction method of multipliers are proposed for $\mathcal {H}_{2}$ and $\mathcal {H}_{\infty }$ problems. The presented design algorithms admit the guarantee of stability for the overall closed-loop system by limited model information exchange. Simulation results for large-scale systems show a noticeable improvement in reducing computational complexity and enhancement of privacy compared to the central design approaches.