PolKA: Polynomial Key-based Architecture for Source Routing in Network Fabrics

A lifecycle experience of PolKA: From prototyping to deployment at Géant Lab with RARE/FreeRtr

“…To solve this problem, PolKA was introduced [Dominicini et al 2020], as a RNS-based SR scheme that replaces the integer arithmetic by polynomial arithmetic (Galois field (GF) of order 2 or GF(2)). The immediate benefit was to enable the reuse of commodity network functions based on polynomial arithmetic.…”

“…In this approach, a source adds a route label in the packet header to define the paths along with the network. Segment Routing and PolKA [Dominicini et al 2020] are two new approaches to the source routing paradigm that allow a flow to be routed to a specific topological path while maintaining a per-flow state only at the ingress node to the SR domain.…”

“…In this paper, we take the position of developers that need to deploy emerging programmable protocols (e.g., PolKA [Dominicini et al 2020]) with specific network requirements (e.g., mod operation). In addition, we want to know how to benefit from the open router platform and testbed infrastructure.…”

M-PolKA: Multipath Polynomial Key-Based Source Routing for Reliable Communications

“…SR also reduces the control signaling and latency related to path setup convergence, so that migrating paths is only a matter of changing the state at source or edge nodes 1 . However, traditional multipath SR solutions lack expressiveness, depend on updating a list state in the packet, and do not offer intrinsic mechanisms to react to failures [12].…”

“…In this scheme, at core nodes, the transmission states of the output ports are given by the remainder of the binary polynomial division (i.e., a mod operation) of the route identifier of the packet by the node identifier. M-PolKA is a generalization of our previous work, named PolKA [12]. It redefines the RNS coding representation for multipath trees, in contrast to the single path routing of PolKA.…”