No abstract

Software-defined networking (SDN) is a novel paradigm that out-sources the control of packet-forwarding switches to a set of software controllers. The most fundamental task of these controllers is the correct implementation of the network policy, i.e., the intended network behavior. In essence, such a policy specifies the rules by which packets must be forwarded across the network.This paper studies a distributed SDN control plane that enables concurrent and robust policy implementation. We introduce a formal model describing the interaction between the data plane and a distributed control plane (consisting of a collection of fault-prone controllers). Then we formulate the problem of consistent composition of concurrent network policy updates (short: the CPC Problem). To anticipate scenarios in which some conflicting policy updates must be rejected, we enable the composition via a natural transactional interface with all-or-nothing semantics.We show that the ability of an f -resilient distributed control plane to process concurrent policy updates depends on the tag complexity, i.e., the number of policy labels (a.k.a. tags) available to the controllers, and describe a CPC protocol with optimal tag complexity f + 2. IntroductionThe emerging paradigm of Software-Defined Networking (SDN) promises to simplify network management and enable building networks that meet specific, end-to-end requirements. In SDN, the control plane (a collection of network-attached servers) maintains control over the so-called data plane (the packet-forwarding functionality implemented on switching hardware). Control applications operate on a global, logically-centralized network view, which introduces opportunities for network-wide management and optimization. This view enables simplified programming models to define a high-level network policy, i.e., the intended operational behavior of the network encoded as a collection of forwarding rules that the data plane must respect.While the notion of centralized control lies at the heart of SDN, implementing it on a centralized controller does not provide the required levels of availability, responsiveness and scalability. How to realize a robust, distributed control plane is one of the main open problems in SDN and to solve it we must deal with fundamental trade-offs between different consistency models, system availability and performance. Implementing a resilient control plane becomes therefore a distributed-computing problem that requires reasoning about interactions and concurrency between the controllers while preserving correct operation of the data plane.In this paper, as a case study, we consider the problem of consistent installation of networkpolicy updates (i.e., collections of state modifications spanning one or more switches), one of the main tasks any network control plane must support. We consider a multi-authorship setting [8] where multiple administrators, control applications, or end-host applications may want to modify the network policy independently at the same time, and w...

No abstract

No abstract

No abstract