The Dilemma and Hope of SRv6

Operators have been fighting "pipelining" for years.

MPLS, SDN... We have been exploring ways to realize intelligent and refined operation networks and realize the network-centric dream.

In the cloud era, the cloud has become the center. SD-WAN has emerged. It is becoming increasingly difficult to de-pipeline.

However, the emergence of SRv6 has rekindled hope.

[[414825]]

The dawn of SRv6

With the rise of cloud computing, various applications have been moved to the cloud. The cloud has become a huge siphon pot, and various resources are constantly concentrated in the cloud.

SD-WAN is like an invisible straw extending from the cloud, overlaying on all networks and siphoning all kinds of resources.

For operators, "the cloud is so deep that we don't know where it is, but we can see the traffic increasing."

It has completely become a pipe. We don’t know what users are doing or what is happening in the cloud. We only see the rapid growth of network traffic. We can only rely on deploying DPI to spy on it, which is expensive and not precise.

The advent of SRv6 gives hope for changing this situation:

• SRv6 packets themselves have a very rich programmable space, allowing operators to identify services more finely and perform scheduling based on identification.

The currently popular APN6 (Application-aware IPv6 networking) technology based on SRv6 technology is designed to achieve this kind of identification and scheduling.

In layman's terms, SRv6 allows operators to block the "entrance" and control the pipeline. If you pay, you can take the main road; if you don't pay, you don't know which road to take.

QoS management is actually the management of human nature: what is worrying is not the lack of supply but the inequality of supply.

The more subdivided the levels are, the more obvious the effect of this management will be.

SRv6 makes it possible to "pipeline control cloud".

The above is a beautiful dream.

Currently, the development of SRv6 seems to be in trouble.

Technical and industrial difficulties of SRv6

The national level has vigorously promoted IPv6, laying a good foundation for SRv6.

However, the SRv6 technical standard is progressing slowly and there are differences in header compression technology.

There are two SRv6 RFC standards:

• IPv6 Segment Routing Header (SRH) RFC 8754
• SRv6 Network Programming RFC 8986

The others are still in the draft stage.

The two standards for SRv6 header compression technology are not unified: G-SID and uSID

The slow pace of technical standards and technical divergences are not just technical issues.

Instead, it has caused many small and medium-sized manufacturers to be unable to quickly invest in research and development. Such manufacturers have no influence in IETF and are generally followers of standards.

Without the participation of a large number of small and medium-sized manufacturers, it would be difficult to achieve a prosperous SRv6 situation relying on only three or four large manufacturers.

Earlier, China Mobile released the "Elastic SD-WAN Technology White Paper", which mentioned many solutions, basically requiring the end CPE equipment to support SRv6:

The terminal CPE equipment is the main field of many small and medium-sized manufacturers.

Simplifying the protocol, standardizing technical requirements, and promoting a large number of small and medium-sized manufacturers to participate in SRv6 research and development are important focuses for getting out of the current predicament.

The fact that only three or four large manufacturers can produce SRv6 CPE does not conform to the law of development.

The dilemma of "who is in charge"

The advancement of technology requires the joint efforts of the three major operators.

Harmony leads to win-win, while division leads to loss. 5G fronthaul has already proved this point well.

However, the attitudes of the three domestic SRv6 companies are still confusing, which is also an important reason why many small and medium-sized manufacturers dare not make up their minds to invest.

China Mobile:

• We have vigorously promoted the development of technical standards, led G-SiD, and released a number of SRv6 white papers.

However, SPN, the network in which China Mobile has made the largest investment, does not support SRv6 and has no evolution roadmap for the time being.

The investment in technology is loud, but the investment in real money is small.

China Telecom:

• STN/new metropolitan area network, comprehensively build SRv6 network, and all newly purchased equipment is required to support SRv6.

However, we haven't seen much technical traction from telecom companies. What kind of SRv6 CPE is needed? Does it support 4G compression?

In terms of investment, telecom is the largest and most resolute of the three. It needs to stand out and lead the way in technology and express a more open attitude.

China Unicom:

• A large-scale SR-MPLS network has been built, but will SRv6 follow? This is still a big question mark.

The industry chain hopes to see the three major operators work together to develop SRv6 with a more open attitude and allow more manufacturers to participate.

Finally: Before every technology matures, it will face a difficult period. Only when the problems are solved during the difficult period can it develop smoothly in the future!