C3PO: Computation Congestion Control (PrOactive) - an algorithm for dynamic diffusion of ephemeral in-network services

TL;DR

This paper introduces C3PO, a proactive, distributed algorithm for managing computation congestion caused by ephemeral in-network services, ensuring low-latency, responsive network performance amidst dynamic service creation and destruction.

Contribution

The paper proposes a novel proactive, distributed computation congestion control algorithm tailored for ephemeral in-network services, addressing a new challenge in network resource management.

Findings

The algorithm effectively distributes service load in dynamic network conditions.

It maintains low latency and high responsiveness in the presence of ephemeral services.

The solution is low complexity, adaptive, and scalable.

Abstract

There is an obvious trend that more and more data and computation are migrating into networks nowadays. Combining mature virtualization technologies with service-centric net- working, we are entering into an era where countless services reside in an ISP network to provide low-latency access. Such services are often computation intensive and are dynamically created and destroyed on demands everywhere in the network to perform various tasks. Consequently, these ephemeral in-network services introduce a new type of congestion in the network which we refer to as "computation congestion". The service load need to be effectively distributed on different nodes in order to maintain the funtionality and responsiveness of the network, which calls for a new design rather than reusing the centralised scheduler designed for cloud-based services. In this paper, we study both passive and proactive control strategies, based on the proactive control we further propose a fully distributed solution which is low complexity, adaptive, and responsive to network dynamics.