On Energy Efficiency of Networks for Composable Datacentre Infrastructures

TL;DR

This paper analyzes the energy efficiency of different network architectures in composable datacenter infrastructures, highlighting the benefits of silicon photonics technology at rack-scale.

Contribution

It introduces a MILP model to evaluate optimal resource disaggregation scale and assesses silicon photonics impact on energy efficiency in composable datacenters.

Findings

Silicon photonics improves network energy efficiency at rack-scale

Rack-scale composable DCs are more energy-efficient than pod-scale with current tech

Optimal disaggregation scale depends on network technology and resource distribution

Abstract

This paper evaluates the optimal scale of datacentre (DC) resource disaggregation for composable DC infrastructures and investigates the impact of present day silicon photonics technologies on the energy efficiency of different composable DC infrastructures. We formulated a mixed integer linear programming (MILP) model to this end. Our results show that present day silicon photonics technologies enable better network energy efficiency for rack-scale composable DCs compared to pod-scale composable DCs despite reported similarities in CPU and memory resource power consumption.