Abstract
Legacy network systems and protocols are mostly static and keep state information in silo-style storage, thus making state migration, transformation and re-use difficult. Software-Defined Network (SDN) approaches in unison with Network Function Virtualization (NFV) allow for more flexibility, yet they are currently restricted to a limited set of state migration options. Additionally, existing systems and protocols are mostly tailored to meet the requirements of specific application scenarios. As a result, the protocols cannot easily be adapted to novel application demands, organically growing networks, etc.
Impeding the sharing of networking and system state, along with lacking support for dynamic transitions between systems and protocols, severely limits the ability to optimally manage resources and dynamically adapt to a desirable overall configuration. These limitations not only affect the network performance but also hinder the deployment of new and innovative protocols as a hard break is usually not feasible and thus full support for legacy systems is required.
On the one hand, we propose a generalized way to collect, store, transform, and share context between systems and protocols in both the legacy Internet as well as NFV/SDN-driven networks. This allows us to share state information between multiple systems and protocols from NFs over BGP routers to protocols on all layers of the network stack.
On the other hand, we introduce an architecture for designing modular protocols that are built with transition in mind. We argue that the modular design of systems and protocols can remove the key limitations of today’s monolithic protocols and allow for a more dynamic network management.
First, we design and implement a Storage and Transformation Engine for Advanced Net- working context (STEAN) which constitutes a shared context storage, making network state information available to other systems and protocols. Its pivotal feature is the ability to allow for state transformation as well as for persisting state to enable future re-use.
Second, we provide a Blueprint for Switching Between Mechanisms that serves as a framework and guideline for developers to standardize and ease the process of designing and implementing systems and protocols that support transitions as a first order principle.
By means of experimentation, we show that our architecture covers a diverse set of challenging use cases in legacy systems—such as Wireless Multihop Networks (WMNs)—as well as in NFV/SDN-enabled systems. In particular, we demonstrate the feasibility of our approach by migrating state information between two instances of the PRADS NF in a virtualized Mininet environment, and show that our solution outperforms state of the art frameworks that are specifically built for NF migration. We further demonstrate that a dynamic switch between WMN routing protocols is possible at runtime and that the state information can be reutilized for bootstrapping novel protocol modules, thus minimizing the control overhead.