Optical Networks for Composable Data Centers

Composable data centers (DCs) have been proposed to enable greater efficiencies as the uptake of on-demand computing services grows. In this article we give an overview of composable DCs by discussing their enabling technologies, benefits, challenges, and research directions. We then describe a network for composable DCs that leverages optical communication technologies and components to implement a targeted design. Relative to the implementation of a generic design that requires a (high capacity) dedicated transceiver on each point-to-point link on a mesh optical fabric in a composable DC rack, the targeted design can significantly reduce capital expenditure (by up to 34 times) because fewer transceivers are used. This is achieved with little or no degradation of expected performance in composable DCs.

Network Topologies for Composable Data Centres

Suitable composable data center networks (DCNs) are essential to support the disaggregation of compute components in highly efficient next generation data centers (DCs). However, designing such composable DCNs can be challenging. A composable DCN that adopts a full mesh backplane between disaggregated compute components within a rack and employs dedicated interfaces on each point-to-point link is wasteful and expensive. In this paper, we propose and describe two (i.e., electrical, and electrical-optical) variants of a network for composable DC (NetCoD). NetCoD adopts a targeted design to reduce the number of transceivers required when a mesh physical backplane is deployed between disaggregated compute components in the same rack. The targeted design leverages optical communication techniques and components to achieve this with minimal or no network performance degradation. We formulate a MILP model to evaluate the performance of both variants of NetCoD in rack-scale composable DCs that implement different forms of disaggregation. The electrical-optical variant of NetCoD achieves similar performance as a reference network while utilizing fewer transceivers per compute node. The targeted adoption of optical technologies by both variants of NetCoD achieves greater (4 - 5 times greater) utilization of available network throughput than the reference network which implements a generic design. Under the various forms of disaggregation considered, both variant of NetCoD achieve near-optimal compute energy efficiency in the composable DC while satisfying both compute and network constraints. This is because marginal concession of optimal compute energy efficiency is often required to achieve overall optimal energy efficiency in composable DCs.

Disaggregation for Energy Efficient Fog in Future 6G Networks

We study the benefits of adopting server disaggregation in the fog computing tier by evaluating energy efficient placement of interactive apps in a future fog 6G network. Using a mixed integer linear programming (MILP) model, we compare the adoption of traditional server (TS) and disaggregated server (DS) architectures in a fog network comprising of selected fog computing sites in the metro and access networks. Relative to the use of TSs, our results show that the adoption of DS improves the energy efficiency of the fog network and enables up to 18% reduction in total fog computing power consumption. More instances of interactive fog apps are provisioned in a fog network that is implemented over a network topology with high delay penalty. This ensures that minimal delay is experienced by distributed users. Our result also shows that the proximity of fog computing sites such as metro-central offices and radio cell sites to geo-distributed users of interactive fog applications make them important edge locations for provisioning moderately delay sensitive fog apps. However, fog applications with more stringent delay thresholds require in situ processing at directly connected radio cell sites or at the location of the requesting users. Finally, we propose a heuristic for energy efficient and delay aware placement of interactive fog apps in a fog network which replicates the trends observed during comprehensive analysis of the exact results obtained by solving the MILP model formulated in this paper. Our results and proposed MILP and heuristic provide a good reference and tool for fog network design and deployment.