Author(s): Ed Doe, Vice President and General Manager, Switch and Fabric Group, Intel Corp.
Use SRv6 to make 5G networks more flexible
Edge computing, network slicing, and low-latency services are long-promised 5G features that have only just been enabled by the transition to autonomous 5G networks. For this to happen at scale, one of the options for 5G and next-gen networks is the benefit of moving from session-based packet routing to IP-based packet routing.
To show how this can be done using Segment Routing over IPv6 (SRv6), SoftBank, Japan’s first mobile network operator (MNO) to offer 5G services, has collaborated with Intel and Arrcus to demonstrate its vision of a SRv6 mobile user plane (SRv6 MUP) network based.
A proof of concept (PoC) of this new SRV6 MUP technology is on display at Intel’s virtual booth at Intel MWC Barcelona “Built for Wonderful” webpage.
Improved legacy circuit-switching architectures
Legacy mobile networks were built using a packet-switched network overlaid on a session-based circuit/network switch. Moving to a packet-only network has the potential to provide the flexibility to support multiple access edge computing (MEC) for user planes closer to the customer, reducing network latency .
Packet-switched networks are also advantageous for enabling network slicing that requires the orchestration of packet flows across the radio access network (RAN), core network, and transport network.
By creating an SRv6 segment routing domain (see Figure 1) within the 5G network specified by 3GPP, the SRv6 MUP can provide the environment for these 5G features — and this is shown in the SoftBank/Intel/Arrcus demo MWC.
SRv6 provides central network control
SRv6 segment routing technology is based on the IPv6 forwarding plane and provides software-defined networking features including traffic engineering and traffic direction.
An SRv6-capable network operating system (NOS) running on switches and routers in the network creates the SRv6 domain where packets are forwarded based on configured network segments to avoid congestion. A packet enters an SRv6 domain through an ingress node and exits through an egress node. Between the two, the packet is transmitted by transit nodes along a predefined path of up to 16 segments. These nodes can be switches, routers, or other network devices based on source routing. The segments can be modified by the controller according to the dynamics of the network. In this way, SRv6 can be used to direct traffic in certain parts of the network.
SRv6 powers the mobile user plane
The solution obtained is detailed in the MWC Demo Video. As shown in Fig. 2, the SRv6 network domain is created using the Arrcus ArcOS NOS. ArcOS runs on a switch using the Intel® Tofino™ Programmable Ethernet Switch ASIC to provide the GTP-U MUP Gateway (MUP-GW) to SRv6 translation for packet processing and forwarding to the 5G core. ArcOS also runs on the MUP-Controller (MUP-C) server powered by a 2nd Generation Intel® Xeon® Scalable processor and MUP provider edge (MUP-PE) switches. MEC traffic is simulated using a 2nd Generation Intel Xeon Scalable processor-based MEC server (MEC server).
The MWC PoC demo simulates performance in an SRv6 network with the MUP controller disabled and enabled to show performance differences. Without the MUP, the packet passes hop-by-hop through the network according to the commands it receives at each hop. This adds latency to the network. But when MUP is enabled, traffic is transmitted more efficiently along network segments with low latency. This is demonstrated by looking at the data streams in the PoC video.
Summary — Advancing Large-Scale Distributed Computing for 5G and Beyond
SRv6 MUP is one of the technology options that enables large-scale MEC-based distributed computing to support low latency and improve network agility for network slicing for 5G networks and beyond. The MWC PoC demo shows how Softbank’s SRv6 MUP network design can be used to inject IP routing into a 5G network to add the flexibility, efficiency, and throughput needed to improve large-scale distributed functionality at scale. SRv6 help.
For more information, see SoftBank’s press release: SoftBank Corp. succeeds in developing SRv6 MUP technology to implement MEC and network slicing at an affordable price on a large scale
Notes and Disclaimers:
- Performance varies with use, configuration and other factors. Learn more about www.Intel.com/PerformanceIndex
- Performance results are based on testing performed on the dates indicated in the configurations and may not reflect all publicly available updates. No product or component can be absolutely safe. Your costs and results may vary.
- Intel technologies may require enabled hardware, software, or service activation.
- Intel does not control or audit third party data. You should consult other sources to assess accuracy.
- ©Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and marks may be claimed as the property of others.