Synergistic network operations

Modern communication networks have multiple ongoing operations that jointly bring about the services provided by the network. In the interests of modularity and design simplicity individual operations correspond to very specific tasks and are optimized for them. However, design and procedural choices made for one operation can potentially impact the performance of another. A classic case in point is when overlay routing decisions disrupt an ISP's intra-domain traffic engineering. Synergistic network operations imply that individual operations cooperate advantageously to enhance the overall efficiency of the network.;The focus of this dissertation is to advocate a holistic paradigm for network operations wherein we jointly optimize design and operational decisions of multiple network operations for maximum global network efficiency. We demarcate two elementary themes for synergistic network operations: (a) network operations should seek to enhance the efficacy of other operations where doing so increases global network efficiency and (b) network operations should be cognizant of and seek to mitigate any disruptive effects on other operations. Ancillary to both these themes is modeling the interaction between different network operations. This is an important first-step towards understanding the opportunity cost of not having jointly optimal design and operation. It also serves to motivate solutions. This dissertation seeks to identify opportunities and outline approaches for collaboration across network operations that exist in practice. Such gracefully collaborating operations might reside at different layers of the protocol stack or at the same layer. This dissertation is inspired by the need to move beyond simple rule-of-thumb principles and best-practices to a more scientific basis for forming a synergistic coalition of network operations. A fundamental challenge in building this coalition is to do so without compromising the inherent deploy-ability, configurability, modularity and manageability of cross-layer and uni-layer network operations.