Optical networks are opening up
By Teresa Monteiro, director of solutions marketing, network software and automation at optical networking company Infinera
Open optical networking is a network architecture where optical signals generated by transceivers (i.e., transponders or coherent optical pluggables) from multiple equipment vendors can run over a line system from a different vendor. This multi-vendor architecture provides communications service providers (CSPs) with more choice, expedites the adoption of optical innovation, and contributes to improved network economics.
Open optical networking has been a topic of discussion within academia, standards organizations, and industry bodies for a couple of decades. However, recent market dynamics and technological advancements have served as a catalyst to accelerate its adoption in real-life networks.
The supply chain shortages encountered during COVID lockdowns underscored how important it is for a CSP to open up its optical networks to multiple vendors and secure second sources for its networking devices. The market introduction in recent years of 400G ZR/ZR+/XR coherent optical pluggables from a range of suppliers also continued to drive the adoption of open optical networks. This is because optical pluggables from one vendor can be directly installed into third-party routers and then connected over yet another vendor’s DWDM optical line system. This disaggregated multi-vendor architecture, known as IP over DWDM (IPoDWDM), is an alternative to connecting routers to the optical line network using traditional transponders and has benefits including saving power, space, and cost.
But CSPs that evaluate introducing new optical suppliers in their incumbent networks are often concerned about managing a disaggregated network efficiently and do not accept a degradation in their operational experience when compared to single-vendor networks. As multi-vendor, multi-platform optical networking becomes standard practice, and coherent optical pluggables are deployed alongside transponders, it is essential to address the operational challenges associated with disaggregation to facilitate its adoption and scaling.
In a multi-vendor open network, CSPs need to visualize all equipment, end-to-end topology, and all services that run over it; they must monitor and optimize network performance and troubleshoot issues that may arise regardless of equipment vendors. CSPs also need to configure the optical layer to provide new end-to-end services across multiple vendors.
Traditional network management systems, which are vendor specific and apply only to specific hardware platforms, don’t fit the bill.
A New Management and Control Approach for Open Optical Networks
The best way to consistently manage all network elements in an open optical network is to adopt a vendor-agnostic software automation solution, often referred to as an optical domain controller. These controllers abstract the equipment’s proprietary details to a management layer above it, regardless of the optical vendor, removing operational challenges associated with open optical.
Standards bodies and industry forums such as IETF, ONF, and OIF continue to work diligently on concepts and architectures for open network management and control, specifying a programmable software control layer that abstracts network hardware using common vendor-agnostic data models and enables network automation via standard open application programming interfaces (APIs).
Using common data models and standard APIs, an optical domain controller can control transponders and coherent pluggables alike, as well as optical line systems, abstracting vendor-specific network details and exposing their functionality through machine-to-machine northbound APIs toward the back office, a hierarchical controller, or other automation applications.
By enabling open optical networking, optical domain controllers help CSPs more rapidly deliver greater network capacity at a better cost.
Addressing a Comprehensive Range of Automation Use Cases
Additionally, optical domain controllers enable a comprehensive range of use cases that service providers can incrementally rely on to simplify their network operations, maximize their return on infrastructure investment, and cope with the industry’s aging workforce.
The primary, foundational application of an optical domain controller is to discover all devices equipped in a multi-vendor network and how they connect to each other, along with the services that operate over them. With this discovery, the optical controller becomes a single pane of glass to the multi-vendor environment: it provides end-to-end network topology and service visualization, along with monitoring performance and alarms from all network devices (transponders, pluggables, and line systems) and corresponding services. Additionally, the optical domain controller expedites troubleshooting by correlating alarms and isolating faults, regardless of vendor.
The creation of new services across an open optical network comes next in terms of automation complexity. This is a task worth automating, as it will likely speed up service delivery. An optical domain controller equipped with path computation element functionality is capable of routing, in real-time, new digital and optical services leveraging up-to-date resource status – i.e., it can route a new 100 GbE demand over an existing wavelength, as well as route a new wavelength across the network. Provisioning the endpoints of a new wavelength may not be a very challenging task for this type of software solution, as protocols based on simple standard data models like OpenConfig are available to support vendor-agnostic configuration of transponders and pluggables. But when routing a new wavelength across a line system, a best-in-class optical controller needs also to be capable of estimating optical performance and validating the feasibility of transmitting a particular modulation format along a specified path. Additionally, the controller needs to handle the configuration of all cross-connections in the optical line system irrespective of vendor, a non-trivial task that typically requires a hierarchical control architecture.
Restoration in open networks is an example of a yet more advanced domain controller use case. It relies on the functionality described above: the controller’s ability to route digital and optical services using real-time network information to redirect and recover traffic on the fly when there are transmission or equipment failures, ensuring high reliability in open optical networks.
An optical domain controller can also be the right platform for the development and deployment of leading-edge automation applications based on data analytics as well as artificial intelligence and machine learning (AI/ML), providing a new level of awareness over network patterns and behaviors. These applications utilize the large amount of data that modern network devices expose through streaming telemetry to generate actionable insights. These insights can be leveraged to dynamically and proactively allocate capacity where necessary, predict degradation, and avoid traffic impacts through preventive maintenance. AI/ML applications can optimize network performance, maximize the utilization of deployed resources, and reduce power consumption. With these capabilities, CSPs can enhance their competitiveness not only by meeting increasingly high customer expectations but also by offering innovative services, with more stringent service-level agreements or shorter delivery times.
Through extensive back-office integration, the optical domain controller offers the means to automate and link multiple workflows and support business processes. This includes interworking with inventory, ordering, or fulfillment systems.
Software Matters When Embracing Vendor Diversity
Today’s vendor-agnostic optical domain controllers support seamless deployment and operation of open optical networks, removing operational challenges introduced by disaggregation.
Optical domain controllers enable CSPs to embrace vendor diversity and reap its benefits, gaining more choice and faster access to innovation, improving network economics, and strengthening supply chain. Additionally, these software solutions support a growing number of network automation use cases, from day-to-day network operation to business process integration, including with the use of analytics and AI/ML to improve network resiliency, optimize investments, and create value for CSPs.
Above, Teresa Monteiro, director of solutions marketing, network software and automation, Infinera
