Disaggregation for Improved Efficiency in Fog Computing Era

TL;DR

This paper demonstrates that using disaggregated servers at the near-edge of telecom networks can significantly reduce the number of fog nodes needed at the far-edge, improving efficiency in fog computing systems.

Contribution

It introduces a MILP model to optimize fog node placement with disaggregated servers, showing potential for up to 50% reduction in far-edge fog nodes.

Findings

Up to 50% reduction in far-edge fog nodes.

Increased average hop count and network traffic.

Marginal rise in network power consumption.

Abstract

This paper evaluates the impact of using disaggregated servers in the near-edge of telecom networks (metro central offices, radio cell sites and enterprise branch office which form part of a Fog as a Service system) to minimize the number of fog nodes required in the far-edge of telecom networks. We formulated a mixed integer linear programming (MILP) model to this end. Our results show that replacing traditional servers with disaggregated servers in the near-edge of the telecom network can reduce the number of far-edge fog nodes required by up to 50% if access to near-edge computing resources is not limited by network bottlenecks. This improved efficiency is achieved at the cost of higher average hop count between workload sources and processing locations and marginal increases in overall metro and access networks traffic and power consumption.