Once loved, monolithic backbone routers will see an inevitable parting of ways between their data and control planes
The Telecom Infra Project Open Optical and Packet Transport (OOPT) community is in the midst of a Request for Information (RFI) process for Disaggregated Distributed Backbone Routers (DDBR) which they believe is where modern IP transport networks are heading at the service provide edge.
A DDBR is a versatile device that can be deployed in IP core/backbone networks as an IP/MPLS core/edge routers (P/PE routers) or an Internet Gateway router (IGW). Compared to traditional routing solutions from a single vendor, a DDBR solution combines software and hardware from multiple vendors, allowing service providers to break vendor lock-in.
TIP said the RFI has been shaped with technical requirements input from Vodafone, KDDI, BT, Orange, Telefónica, Bell Canada, Telia and MTN to ensure that the RFI process aligns with the real-world needs and expectations of telecom operators. Responses will be evaluated against industry-validated interoperability and compatibility standards, ensuring seamless integration with diverse network environments.
Existing IP backbone routers are predominantly built based on a chassis structure with front access where the Interfaces (NNI/UNI) cards and the control boards are plugged into a common backplane. This model worked well but to build in all the extra redundancy and resilience they ended up with high computing capability, high capacity and high cost. That means larger data centre footprints and finite slots for new services. Also having all the NNI & UNI interfaces centralized into the same chassis is risky if the entire node is lost for something as simple as a misconfiguration or software glitch.
Pluggable shift
TIP points out that with the industry shift happening in the optical pluggable transceivers and the dawn of 400G QSFP56-DD optics, the services providers need to replace the current installed base to higher capacity, more compact dimensions, flexible thermal management ports which enable supporting higher capacity links with optimum port density per RU.
In addition to the 10G/25G/40G UNI interfaces, 100G and 400G UNI and NNI interfaces need to be supported to get the outmost of the interface capacity through the backplane bus. As a result, telcos need a new approach to protect their IP backbone investments.
Hyperscalers overcame these issues by move to spine and leaf-based architectures to disaggregate the control plane from the data plane, which instantly solves the dependency on the router switching capacity and the number of interfaces. TIP points out that with less risks compared to entirely migrating the service to a new chassis in the scale-up model, upgrades in Spine & Leaf-based architecture is imposing less operational risk for instance, you just need to connect a new spine router to increase the overall system capacity and connect a Leaf router to increase the number of interfaces. This is where DDBRs will play out.
As part of the RFI, each vendor’s response will be evaluated against a threshold-based criteria, gauging the completeness of the solution against the comprehensive set of requirements outlined in the Detail Technical Requirements document. Submissions are due by 2 February.