The original formulation of the coded caching problem assumes that the file requests from the users are synchronized, i.e., they arrive at the server at the same time. Several subsequent contributions work under the same assumption. Furthermore, the majority of prior work does not consider a scenario where users have deadlines. In our previous work we formulated the asynchronous coded caching problem where user requests arrive at different times. Furthermore, the users have specified deadlines. We proposed a linear program for obtaining its optimal solution. However, the size of the LP (number of constraints and variables) grows rather quickly with the number of users and cache sizes. In this work, we explore a dual decomposition based approach for solving the LP under consideration. We demonstrate that the dual function can be evaluated by equivalently solving a number of minimum cost network flow algorithms. Minimum cost network flow algorithms have been the subject of much investigation and current solvers routinely solve instances with millions of nodes in minutes. Our proposed approach leverages these fast solvers and allows us to solve several large scale instances of the asynchronous coded caching problem with manageable time and memory complexity.