Under the background of cloud network convergence era, flexible and agile network service capability directly affects the competitiveness of operators. The emergence of SRv6 is a huge innovation. Combined with SDN technology to enable programmable networks, it provides the soil of innovation, for basic network services and value-added network services, in the cloud network era.
SRv6 inherits all the advantages of SR-MPLS and makes major improvements. By unifying the network forwarding plane to IPv6, it can establish a cross-domain forwarding path as long as the IPv6 routing is reachable, while the dedicated MPLS forwarding plane is no longer required; By defining the functional instruction information, it makes packet forwarding and processing programmable, from the perspective of macro network. SRv6 technology uses the 128bit IPv6 address as its Segment ID (SID). With its minimal programmable features, SRv6 enables any access and any join, flexibly meets various network service requirements, and provides higher reliability and scalability, which will become the core technology of the next generation IP bearer network.
There are two main challenges which SRv6 technology is currently facing in the actual network deployment: Firstly, SRv6 packet cost is big, which leads to lower link bandwidth utilization. For example, the bandwidth utilization will be only about 60%, if the packet length is 256bytes while it has 8-layer SID; Secondly, it brings higher requirements to chips when processing SRv6 packet. It is very difficult for the existing network equipment to support the replication and operation of the deep SRH head. Moreover, it will reduce the processing efficiency for the current chips to process 128bit SID. To solve the problems in standard SRv6 technology mentioned above, it is necessary to compress and optimize the SRH head of standard SRv6, by adopting simpler and shorter G-SID such as the typical 32bit length. We can use 32bit G-SID to represent and replace the standard 128bit SID in SRH head, to optimize the SRH head encapsulation cost of standard SRv6, to realize SRv6 efficiency optimization while supporting SRv6 function, to support the large-scale deployment of SRv6 technology in providers’ networks.
In order to promote the technology solutions and products of SRv6 and SRv6 head compression optimization completable as soon as possible, China Mobile has launched an overall open-source solution for SRv6 technology jointly with a number of partners, which is composed of controller, operating system and forwarding plane, aimed at promoting the SRv6 technology system landing, to accelerate the process of cloud network convergence. In view of the SRv6 protocol system, this project proposes a complete combination of forwarding devices, by integrating three new open-source mainstream platforms, which takes G-SRv6 as the core network protocol, composed of ODL-based controller, SONiC-based operating system, and programmable open-source language. This project innovatively builds an end-to-end G-SRv6 open-source system for IP bearer network, so as to realize the revolutionary simplification and unification of network architecture, the automation of service deployment, and the overall solution of core routing system for cloud network convergence.