This paper proposes a new two-phase framework to solve a multi-target rendezvous problem. The first phase of the framework constructs a multi-layer elementary solution database composed of two-impulse and three-impulse optimal transfer solutions between two targets. The second phase formulates and solves a combinatorial optimization problem to find the optimal rendezvous targets and visiting sequences using the elementary solutions. The first phase finds local minima for two-impulse single lambert rendezvous. And then, to find the three-impulse solution, the grid-search and gradient-based algorithm is applied sequentially with the two-impulse solution as the initial guess, reducing the computational cost. The set of local minima for the various operating modes found in this way can be used as an inner-loop to establish a multi-layer elementary solution to determine the optimal visit sequence to multi target rendezvous. The numerical examples were carried out to verify the validity of the corresponding methodology.