When the Rubber Hits the Road: Deploying, operating and maintaining disaggregated networks is simpler than you think

This is a reasonable concern. Constructing a router cluster composed of dozens of white boxes that have the same capacity as one single chassis, or using different vendors for software and hardware sounds like it would make any operation more complex. You may be inclined to believe that, one box, one vendor, one interface is preferable to using multiple boxes, vendors or interfaces. But is this belief true?

Deploying, operating and maintaining a multi-vendor solution with many white boxes and multiple hardware and software combinations is ACTUALLY simpler than a traditional, single vendor, monolithic chassis. Really! This is because we make sure that every element – hardware, software containers, ODM vendors – operates in defined patterns, ensuring predictability and visibility while supporting easy automation.

For Legacy Networks Operations Matters

The more networks grow, the more complex they become. In the past 30 years, networks have increased their use of hardware equipment, nodes and trunks, making them difficult to install, expand, manage and maintain. Network complexity makes us blind when we need to increase capacity, launch a new service or fix a network issue. This blindness is costly – impacting the network’s operational cost or OpEx. For every dollar of network CapEx, operators can spend between $3 to $5 in OpEx.

The traditional monolithic chassis – what has served as the foundation for legacy networks – adds extra operational complexity, including:

A “black box” approach with a lack of granular visibility and control
Monolithic software as a potential source of compatibility issues and bugs, like during upgrades
Legacy interfaces not adapted to operational tasks (CLI, SNMP, proprietary models)

Network operators have addressed this complexity by working only with a specific vendor that oversees the entire process of deployment, support, and maintenance including all liabilities. In other words, it’s more complex, but the related risks are passed over to the vendor.

Disaggregated Networks: The Concerns

The disaggregated network approach also raises specific questions around operations, such as:

Certification, planning and installation – how can you be efficient when you need to certify, plan and install a high number of boxes and cables and deploy power and cooling systems across multiple sites?
Setup and initial provisioning – how do you combine hardware and software to make a working node, while a chassis comes already built-in?
Life-cycle management – how do you control the management of both the network and service life-cycles – including hardware and software – as new devices are added and still avoid complexity?
Support- with multiple players involved, how do you ensure good problem isolation and avoid fingerpointing between vendors?
OSS/BSS integration – how do you minimize the integration effort required for existing OSS/BSS tools, which were designed to work with traditional network devices?

The ideal answer is an operational model that not only addresses these concerns without adding extra complexity but also removes some of the inherent complexity that came with operating traditional networks.

A New Operational Model

This new operational model is what DriveNets built from scratch and optimized over multiple tier-1 operator deployments. DriveNets leads all operational aspects such as inquiries, feature requests, design and planning considerations and critical network-down troubleshooting, regardless of the root cause of a malfunction.

Let me clarify some key points:

1. Hardware certification, rack/site planning and installation are simpler

The end-to-end system certification process is managed by DriveNets, including hardware at box and rack levels. For any network or routing function, DriveNets uses only 2 hardware building blocks (NCP for packet forwarding and NCF for fabric). Once qualified for one network domain (core, edge, other), they are qualified for all network domains. Due to their size (2 Rack Units), installing standard, small boxes – even in high quantities – is simple and fast since there’s no need for heavy machinery or a specific skill set. Installing cables is also easier, since they span over multiple racks and not in a small patch area like in a chassis setup. DriveNets provides a detailed plan to connect the different ports, avoiding the possibility of errors during setup.

2. Built-in automation and zero-touch provisioning delivers fast time-to-service

For each cluster size, DriveNets Network Orchestrator (DNOR) makes sure the right software – BaseOS and DNOS – is installed automatically and securely to build a Network Cloud entity. This is important not only for the initial deployment stage but also when adding capacity – and more NCPs and NCFs. Automated provisioning and remote testing avoids the issue of undocumented pre-production failures that happen with a traditional chassis, offering fast time-to-service.

3. No more complexity of managing many individual boxes

DNOR automates the full life cycle of hardware and software components (firmware, BaseOS, DNOS containers) across nodes, clusters and the entire network. From a user viewpoint, it appears to them like managing one router – just like today – but with full visibility and a selective control of the different software components (as opposed to a monolithic solution). For example, upgrading a 192 Tbps router with a new OS software version would take no more than 30 minutes.

4. One throat to choke support

DriveNets serves as a single point of contact for any unplanned incident (software or hardware), from detection to remediation. This provides a better customer experience and clear responsibility and accountability between partners, offering:

A unique tier-1 technical support number
A common technical support ticketing system (Jira) – available 24/7/365, shared between all parties
A triage system to either engage the right hardware partner(s) or DriveNets as a software supplier
Full transparency of each case status with the engaged SLAs
Granular DNOR’s monitoring system to ease problem isolation

5. OSS/BSS integration won’t take ages

DriveNets offers different paths to simplify OSS/BSS integration, such as:

DNOS Yang models can be converted to the OpenConfig standard for each northbound interface (gNMI, Netconf)
DriveNets Network Cloud CLI commands can be translated to the equivalent legacy CLIs
Rest APIs are exposed through DNOR for provisioning and configuration

DNOR is the most comprehensive option as it delivers automated provisioning, operation and health monitoring, offering full GUI-based visibility and advanced control of the entire solution.

The Simplicity of Complexity

In nature, a flock of birds reflects an example of a complex adaptive system (called “flocking”). A flock of birds can appear to move in unison, shifting direction dramatically as if the flock shared one mind or was led by a single bird. In fact, the flock has no real leader or mind-melding abilities but is composed of many individual elements all moving in the same direction together.

This is how operating a disaggregated network looks like.

Disaggregated networks are actually easier to operate and manage than legacy networks. This same “flocking” approach takes place, turning the various different embedded hardware-centric functions of a traditional router into separate manageable software-driven planes or components, while appearing as a single-vendor integrated chassis. As DriveNets takes the full, end-to-end responsibility of building and maintaining the entire solution, network operators can enjoy a safe and simple path for deploying disaggregated networks, truly taking their network to the next level without all the complexity and concerns.