Service Provider Routing

NOS – Network Operating System

As networks have become more software-centric, the Network Operating System (or NOS), has become a key enabler of network and service innovation.

There are several approaches to NOS ranging from monolithic to fully disaggregated.

Monolithic NOS: In this approach, the NOS vendor is the same as the hardware provider. A Cisco router will use only Cisco IOS NOS, a Juniper router only JunOS, and a Nokia router only Nokia SR OS. This limits innovation as it ties hardware and software innovation to the same path.
Pseudo disaggregated: This is a NOS that comes from the hardware vendor but can, theoretically, run on another type of hardware. Cisco’s IOS XR, for instance, runs on Cisco routers that incorporate the Cisco Silicon One ASIC, but supposedly can run on third-party hardware too. This is also the case with Nokia’s SR Linux.
Disaggregated NOS: Developed by software companies, this type of NOS runs on COTS white boxes. DriveNets Network Operating System (DNOS) is one such NOS. DNOS allows the best flexibility in choosing the right hardware at the right cost, harmonizing the hardware across network domains and, most importantly, accelerating innovation and creating new services and revenues streams.
Open-source NOS: This is similar to disaggregated NOS, but as a free-to-use, open-source option. A notable example is SONiC. While their benefits are obvious, open-source NOSs lack necessary functionality for a many use cases and require intensive tailoring and in-house software capabilities; hence, they are mostly used by large hyperscalers.

Did you know that a NOS can feature additional capabilities such as virtualized distributed networking and native “cloudification” of the network?

Learn more here

Optical Transport

Optical transport is key for enabling the ever-growing demand for capacity. For decades, technologies like dense wavelength division multiplexing (DWDM) and optical transport network (OTN) allowed operators and hyperscalers to grow their infrastructure capacity without the need for further investment in physical infrastructure (e.g., laying fibers underground or undersea).

This created a separately managed optical domain that includes optical nodes such as reconfigurable optical add-drop multiplexers (ROADMs), repeaters and amplifiers across the network.

More importantly, it created significant optical overhead for any edge site. This includes terminal multiplexers, which incorporate transponders and muxponders that are used to convert signals coming from the data source (e.g., the router, switch, or base station) into “colored” signals on a specific wavelength.

Developments: Major developments in coherent optics, mainly the introduction of next-generation ZR and ZR+ optics, have enabled the collapsing of the entire terminal multiplexer/transponder/muxponder edge-site shelf into optical pluggables. Such pluggables go directly into the switch or router at the edge site.
Benefits: This brings huge benefits in terms of CAPEX and OPEX (derived from reduced floor space, power, A/C etc.). Instead of building an entire rack, operators and hyperscalers can use ZR/ZR+ optical pluggables (including 100Gbps ZR, 400Gbps ZR+, 800Gbps ZR+ and more) plugged directly into an edge router.
Disaggregated network architecture: In a monolithic architecture, those ZR/ZR+ optics can only be sourced from the router vendor, which reduces the level of flexibility in vendor selection and, in some cases, prevents end-to-end optical planning and management. With a disaggregated router architecture, any optical ZR/ZR+ module can be used, maintaining the benefits of an end-to-end optical strategy.

How does DriveNets support the converging of optical and IP networks with open ZR/ZR+ optical interfaces here?

Learn how

Routing

Introduced by Cisco in 2011, segment routing (SR) is a flexible routing method within a single autonomous system (AS) or network domain. Having since gone through IETF standards processes, it has been adopted by a few network operators worldwide, and implemented by most networking vendors.

Breaks network topology into segments: Each segment is represented by a segment identifier (SID). The SID can be represented as an MPLS label (aka SR-MPLS) or as an extension header in the IPv6 packet (aka SRv6). Using a collection of SIDs imposed on the packet, the packet can then be “steered” through the network based on specific policies.
Simplifies network operations: SR can simplify network operations, especially for a large-scale network, by imposing predefined and dynamic traffic routes without requiring extensive state maintenance. It does this while allowing the use of existing MPLS infrastructure via SR-MPLS, making it easier to adopt for most service providers.
Segment routing challenges: Transitioning from traditional MPLS networks to SR-MPLS is still a complex process that most likely will require a hybrid solution during the migration period. Furthermore, a transition from traditional MPLS to SRv6 can be even more challenging since shifting thousands of already used services based on labeled IPv4 is extremely complicated.

Segment routing, and especially SRv6 if deployed correctly along with a more modern architecture, offers service providers a robust, scalable, and cost-effective solution for modern traffic engineering. It dramatically reduces network operations efforts and supports rapid service delivery.

Industry Events

Disaggregation Meets Optics Innovation at OFC

At OFC 2024, DriveNets' Calin Miculescu P.Eng demonstrates the future of networking,...

Blog

What is DriveNets Network Cloud?

Evolution is inevitable in any market, as the competitive environment and changes in market demands accelerate evolution. Some players, such...

White Papers

Which Network Architecture Is Right for You

This white paper examines how service providers should evaluate next-generation architecture for multi-service routing applications. There is a tendency to...