Access Official Technical PDF

Share Summary:LinkedInEmail

A modern metro network architecture must do more than carry traffic between access networks and the core. As 5G services introduce varied bandwidth, latency and processing needs, transport and computing resources need to be planned together. The METRO-HAUL project explored a framework for combining optical transport, packet switching, distributed compute and storage, and network orchestration across metro networks.

Rather than prescribe one design for every deployment, the approach defines architectural options and methods for comparing them. This is relevant to engineers and operators weighing how to scale capacity, place edge resources and manage infrastructure costs and energy use across different geographic and traffic scenarios.

Why metro networks need an application-aware design

Metro networks sit between heterogeneous fixed and wireless access clients and the photonic core. They aggregate traffic from residential, enterprise and mobile services, but those services do not necessarily have the same requirements. Some need substantial transport bandwidth; others may benefit from processing or storage closer to the point where traffic enters the network.

The project therefore frames metro infrastructure as a combination of network and IT resources. Its design objective is to support growing traffic without assuming that fiber deployment must expand at the same rate. It also considers how application-specific requirements can influence the allocation of bandwidth, computing and storage.

Two edge-node roles in the metro network architecture

The architecture distinguishes between Access Metro Edge Nodes (AMENs) and Metro Core Edge Nodes (MCENs). Both combine optical networking and compute capabilities, but they serve different positions and functions in the network.

  • AMENs aggregate traffic nearer to access networks. Their distributed compute and storage can support services with particular bandwidth or IT requirements.
  • MCENs operate closer to the metro-core side. They are designed for higher optical throughput, with compute and storage roles that may include caching or more centralized processing and control.

A node can be understood as three interconnected building blocks: an optical system, a packet-switching fabric and a compute or storage platform. The optical system includes the line system and transceivers; the packet fabric aggregates and forwards access traffic; and the compute platform hosts virtual network functions (VNFs). Interconnecting these parts enables network services to use resources across multiple technology layers.

Planning optical transport and node options

There is no single optical-node configuration suited to every metro environment. Candidate designs examined in the project include different approaches to filtering, switching and wavelength management, including filter-based, filterless and semi-filterless architectures. The relevant choice depends on the network scenario and its requirements, including traffic, topology and the distances optical interfaces must cover.

Optical disaggregation is another design consideration. A disaggregated model separates elements that may otherwise be supplied and managed as an integrated system. The project’s evaluation framework considers the potential economic effects of different levels of disaggregation, rather than treating openness as an automatic cost advantage. Architecture decisions need to account for the components and interfaces involved in each option.

For engineering teams, the practical implication is to compare complete configurations, not isolated devices. The optical line system, transceivers, switching fabric and compute resources influence one another. A change in one part of the node can affect how traffic is transported or where service functions can be placed.

Edge computing requires joint network and compute planning

Adding compute to metro locations creates opportunities to place VNFs closer to traffic, but placement must be considered alongside transport capacity and service requirements. The METRO-HAUL methodology examines several planning problems, including data-center dimensioning against latency classes, dynamic service chaining, baseband-unit placement and content delivery network (CDN) cache planning.

These are distinct problems, but they share a central question: where should a function or resource be located, given the needs of the service and the capacity of the network? For example, service-chain planning considers how traffic reaches a sequence of VNFs. Cache planning considers the relationship between storage at edge nodes and the bandwidth needed to deliver content. In both cases, a decision about compute or storage can change transport requirements.

Because demand varies, the framework also models traffic at different timescales. Macro-level inputs describe service types and growth assumptions, while finer-grained models account for daily or weekly patterns and changing service flows. This helps planners evaluate architectures against more than a single static traffic snapshot.

Orchestration and techno-economic evaluation

Coordinating packet, optical and compute domains calls for an orchestration framework. In the proposed model, network control works alongside management of distributed virtualized infrastructure. Services can be represented as combinations of VNFs and associated logical or physical resources, supporting the concept of network slicing across the infrastructure.

Automation is also relevant to changing conditions. The architecture considers how control, management and orchestration can reconfigure bandwidth and VNFs as service requirements change. Open interfaces and common models, including NETCONF/YANG-based approaches, are intended to support interaction between devices, network software and higher-level applications.

To compare candidate architectures, the project identifies network capacity, capital expenditure and energy consumption as key techno-economic measures. Its planning framework uses reference topologies and geographic scenarios, traffic models, planning tools, and cost and power models for optical, packet and data-center components. Together, these inputs provide a structured basis for assessing alternatives. They do not make one design universally preferable: results depend on the scenario and assumptions being evaluated.

Practical implications for network planners

For operators, integrators and engineering teams, the main lesson is to plan metro transport and edge infrastructure as a connected system. Start with the geography, service mix and traffic patterns; then assess optical architecture, node roles and compute placement against the same requirements. Compare capacity, cost and energy together, and consider how orchestration and open interfaces fit the intended operating model.

This approach can make design trade-offs more explicit before deployment. A flexible metro network is not defined by a single technology choice, but by selecting and coordinating suitable optical, packet and compute components for each use case. The result is a more systematic basis for evaluating how metro infrastructure can support 5G services and future changes in demand.

Access Official Technical PDF

About The Author