Until recently, it was believed that long-lived quantum memories were necessary for long-distance quantum communication. However, by using error-correction codes in an efficient way—specifically, by correcting for photon loss—it is possible to transmit quantum information over long distances without quantum memories. For quantum computation, recent architectures for topological quantum computation indicate that the simplest large-scale structure could be memory-less. While a quantum memory may no longer be an essential resource for quantum networks, it could nonetheless be a key device in the development of quantum information technology. However, it is still not clear what benefits a functioning device could bring to quantum information systems, largely due to a lack of detailed models. Recently we have developed a detailed model for a quantum network based on a simple device designed to act as a building block for a full system architecture. The device is based on an optical cavity containing a negatively charged nitrogen-vacancy center in diamond. This model naturally integrates quantum communication with computation, and using this model we can assess quantitatively the costs and benefits of quantum memories. With or without quantum memories, it is necessary for us to preserve quantum information for a long period of time in either communication or computation.