The New SP Network – Resolving Challenges with a New Network Blueprint

While this year’s MWC is in March, let’s take a look now at the state of the telecom service provider (SP) industry. What are its main challenges, goals and opportunities, and how should they be addressed?

Telecom SP market – starting to change

The telecom business is sluggish – this is not new. There are two trends that contribute to that: near-stagnant revenue streams, and growing capital expenditures (CapEx) and operating expenses (OpEx) deriving from the ever-growing demand for data. So far, same old story. But something is starting to change. We are seeing more and more industry leaders taking a clear path towards changing the above equation. Specifically, they want to take care of the most painful part of it – OpEx. In order to do that, there needs to be a fundamental change in the way network infrastructure is built (and planned, operated and maintained). There are two main building blocks that need to be implemented in this new architecture: convergence and automation. These are enabled by additional underlying concepts – specifically disaggregation and AIOps. Let’s touch upon each of these building blocks and underlying concepts…

Convergence – more than one way

At the end of the day, the target of convergence is to reduce the number of network elements, and, even more importantly, the number of SKUs in the network. For that, there are multiple technologies and architectures that can be implemented:

• Optical & routing convergence: The wide availability of pluggable coherent optic modules, specifically with ZR and ZR+ technology, allows operators to collapse the edge of the optical network into the routing elements. This is extremely important in edge sites, in which space and power are limited, and the elimination of optical transponders allows the support of more capacity and more services with the same footprint and power consumption. It’s important to remember, however, that this needs to be implemented within an open ecosystem; this will enable operators to maintain flexibility in choosing best-of-breed vendors in each domain (even if the two domains reside in the same box).
• Unified edge network architecture: Here, we eliminate the silos of the edge and aggregation networks. No more separate edge devices for business and residential access, or for fixed and mobile access. All traffic runs through the same devices (though maintaining logical separation where needed). This leads to a dramatic drop in the number of network elements and even in the number of network sites.
• Protocol stack simplification: This one requires a leap of faith, but the introduction of SRv6 (Segment Routing over IPv6) can lead to a reduction of the number of protocols in the network (IGPs, specifically). It also leads to a reduced number of routing domains and to a significant simplification of planning and operations processes.
• Unified OSS integration: An outcome of the above is a reduction of protocols and northbound interfaces toward the operational support system (OSS). Integration of new services and functionalities becomes much simpler, while enabling very fast time to market (TTM).

Automation – cloud-like operations

Massive automation brings us closer to an autonomous network, in which no command-line interface (CLI) is used. This leads to reduced OpEx and better mean time to recovery/repair/respond/resolve (MTTR), mean time between failure (MTBF), and time to market (TTM). Automation is available throughout the network and service lifecycle:

Planning, setup and testing:

• “Digital twin” – providing a mirror image of the production network in which changes can be tested and validated without interrupting service
• Auto traffic management and capacity planning – allowing a proactive, touchless, ongoing planning process

Deployment and maintenance:

• Auto configuration with zero-touch provisioning (ZTP) – allowing first-time-right configuration with faster TTM
• No CLI usage and full remote-control capabilities – reducing security hazards and human errors

Operations and optimization:

• Auto network inventory – HW & SW call home every ~15 minutes, maintaining a single source of truth regarding network and equipment inventory
• Storm network control – providing a graceful, prioritized restart to prevent a domino effect

Disaggregation – the key enabler

To achieve all of the above mentioned goodies, you’ll need a disaggregated, software-based network infrastructure. Specifically, you’ll need a network operating system (NOS) that was built as a hardware/optics/ASIC-agnostic system, which allows operators to streamline hardware across the entire network. This reduces the number of SKUs in the network and enables true convergence. Moreover, an open system will allow maintaining a best-of-breed approach for optical networks even when utilizing pluggable coherent optics (with no constraints on the routing vendor to use its own optics). This software-based architecture is also the key enabler for simplifying the northbound interfaces towards the OSS layer. Disaggregation, essentially, is a page out of the hyperscalers’ book, and, inherently, is limited to new types of networking vendors. Incumbent routing vendors have some inherent limitations when implementing this new architecture. These include the shifting focus away from telecom and towards enterprise and cloud, their dependency on their lucrative hardware business, and their NOS not being built as hardware-agnostic software.

AIOps – the proof of the pudding

Harnessing artificial intelligence (AI) to simplify network operations is a limitless opportunity, and there are many examples of doing just that. In fact, DriveNets has developed a number of cool tools that can make the life of operations teams much, much simpler.

To learn about those tools, though, you’ll need to stop by our booth (2K65) at MWC next month in Barcelona. Hope to see you there…