DIP Networking

Author : Xiaoliang Zheng, Wei Shao

 Copyright
Author: Xiaoliang Zheng, Wei Shao
Key Contributors: Fenghua Zhao, Yanjun Liu, Lanjun Luo, Ruiqiang Lu
Release Date: 2021-08-20
Issue: 01

Copyright © Huawei Technologies Co., Ltd. 2021. All rights reserved.

No part of this document may be reproduced or transmitted in any form or by any means without prior written
consent of Huawei Technologies Co., Ltd.

Trademarks and Permissions

and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd.
All other trademarks and trade names mentioned in this document are the property of their respective holders.

Notice
The purchased products, services and features are stipulated by the contract made between Huawei and the
customer. All or part of the products, services and features described in this document may not be within the
purchase scope or the usage scope. Unless otherwise specified in the contract, all statements, information, and
recommendations in this document are provided "AS IS" without warranties, guarantees or representations of
any kind, either express or implied.
The information in this document is subject to change without notice. Every effort has been made in the
preparation of this document to ensure accuracy of the contents, but all statements, information, and
recommendations in this document do not constitute a warranty of any kind, express or implied.
 Preface

Author Introduction
Xiaoliang Zheng: Expert in the Router Design Department of Huawei's Data
Communication Product Line. He joined Huawei in 2005 and has served as a
development engineer, architect, system engineer, and maintenance system
engineer. Currently a system engineer in deterministic IP (DIP) networking
technology research, his extensive working experience has given him a
comprehensive understanding of data communication products.

Wei Shao: As a senior datacom documentation engineer in Huawei, she has long
been engaged in developing technical documentation for key features of
datacom products. She participated in the planning of IPv6 series eBooks and is
dedicated to promoting relevant technologies.

About This Book

This book describes the background and value of DIP networking and shows its
diversified application scenarios and promising future. In particular, Chapter 5

i
Preface
explains how DIP networking works. This book aims to give you a deeper
understanding of the technical advantages of DIP networking.

Intended Audience
This book is intended for network planning engineers, network design engineers,
mid- and senior-level managers at service providers and enterprises, and readers
who want to understand cutting-edge IP network technologies. Because DIP
networking involves many network concepts, readers of this book should be
familiar with IP network basics, such as the IP network architecture, Segment
Routing over IPv6 (SRv6), and Time-Sensitive Networking (TSN).

ii
Preface
 Acknowledgments

In writing and publishing this book, we received extensive help and support from
both inside and outside Huawei. We sincerely thank Jinzhu Chen, Meng Zuo,
Zhenbin Li, Zhiqiang Du, Zhaokun Ding, Dawei Fan, Chenxi Wang, Wenjun Meng,
Tao Han, Hongkun Li, Fenghua Zhao, Yue Liu, and other leaders and experts
from Huawei Data Communication Product Line for their guidance and support.
Our thanks also go to Hui Tian, Shujun Han, Danni Ma, and other experts from
China Academy of Information and Communications Technology, who not only
provided valuable technical guidance but also carefully reviewed the book.

This book focuses on the most cutting-edge IPv6 technologies, which are still
evolving and deepening. While we have made significant efforts to ensure
accuracy, there might be omissions or deficiencies in the book. Your comments
and feedback are warmly welcomed.

i
Preface
 Table of Contents

Chapter 1 Overview of DIP Networking......................................................................... 1

Chapter 2 Background of DIP Networking .................................................................... 3

2.1 IP Network Problems ...................................................................................... 3

2.2 Deterministic Transport Required by the Industrial Internet ............. 6

2.3 Deterministic Transport Required by 5G to Enable Numerous

Industries .................................................................................................................. 10

2.4 Emergence of DIP Networking................................................................... 11

Chapter 3 Value of DIP Networking ............................................................................. 12

3.1 Facilitating Evolution of the Industrial Internet ................................... 12

3.2 Advancing All-Industry Access to 5G Networks ................................... 14

3.3 Accelerating New Service Incubation....................................................... 15

Chapter 4 Overall Architecture of DIP Networking ................................................. 16

Chapter 5 Key Technologies of DIP Networking ...................................................... 20

5.1 Edge Shaping ................................................................................................... 20

5.2 Gate Scheduling .............................................................................................. 21

ii
Table of Contents
 5.3 Interval Mapping ............................................................................................ 22

5.4 SRv6 Explicit Path Planning......................................................................... 23

5.5 Dual Fed and Selective Receiving ............................................................. 25

5.6 Summary ........................................................................................................... 26

Chapter 6 Applications of DIP Networking ................................................................ 27

6.1 CENI .................................................................................................................... 27

6.2 Cloudified PLC Project of China Telecom Shanghai ............................ 30

Chapter 7 Future Prospects of DIP Networking ........................................................ 34

iii
Table of Contents
 Chapter 1
Overview of DIP Networking

Deterministic IP (DIP) networking is a new Layer 3 networking architecture.

What makes a network deterministic? A network is considered deterministic if it

provides bounded latency or bounded jitter. As shown in Figure 1-1, bounded
latency means that the latency is less than or equal to the upper bound T. That
is, the latency and jitter each ranges from 0 to T. Bounded jitter, on the other
hand, means that the latency is T and jitter is less than or equal to the upper
bound Δt. That is, the latency ranges from (T – Δt/2) to (T + Δt/2), and the
jitter ranges from 0 to Δt. Based on this, it can be seen that the requirements
for achieving bounded jitter are stricter. This is because the latency for bounded
jitter is less than or equal to the upper bound T + Δ t/2, meeting the
requirements of both bounded jitter and bounded latency.

1
Overview of DIP Networking
 Figure 1-1 Bounded latency and jitter

On a DIP network, an innovative periodic scheduling mechanism is introduced to

the data plane. This reduces microbursts by controlling the forwarding interval of
each data packet at each hop, achieving both bounded jitter and bounded
latency. The control plane uses an efficient path planning and resource
allocation algorithm that does not require flow-by-flow, hop-by-hop timeslot
orchestration, thereby enabling large-scale, expansible, and low-latency DIP
networking.

2
Overview of DIP Networking
 Chapter 2
Background of DIP
Networking

Abstract
Traditional IP networks do not provide deterministic latency during E2E
packet forwarding. However, both the industrial Internet and 5G vertical
industries require networks to provide deterministic transport. It is for
this purpose that DIP networks emerge.

2.1 IP Network Problems

Traditional IP networks, by their very nature, do not provide deterministic latency
during E2E packet forwarding, and bursts of service packets aggravate latency
uncertainty.

3
Background of DIP Networking
Connectionless-Oriented STDM Network
An IP network is connectionless-oriented and implements Statistical Time-
Division Multiplexing (STDM). Figure 2-1 shows how packets are multiplexed.
Packets from different inbound interfaces are first aggregated and then sent
from the same outbound interface. Which packet is scheduled first depends on
the sequence in which they arrive at the outbound interface's queue. Specifically,
packets that arrive first are sent out first.

Figure 2-1 IP statistical multiplexing

In connection-oriented communication, a batch of packets is identified as one

user, and STDM processing is performed based on the user. In connectionless-
oriented communication, in contrast, there is no user involved. To put it another
way, each packet represents a user. As such, each packet gets an equal
opportunity for scheduling.

In synchronous TDM, each user is assigned a fixed time slice and fixed
bandwidth. In STDM, time slices are dynamically divided among users based on
their traffic usage, and the bandwidth used by each user is uncertain and
changeable.

The benefits of connectionless-oriented STDM are fully utilized network

bandwidth and reduced network investment of carriers. The accompanying
disadvantage, however, is that packet forwarding latency is uncertain. As shown
in Figure 2-2, although the forwarding latency is within an acceptable range for
most packets, there will always be a few packets that have unacceptable

4
Background of DIP Networking
forwarding latency due to severe packet bursts. This results in the long-tail
effect.

Figure 2-2 Long-tail effect for packet forwarding latency on an IP network

Service Traffic Bursts Aggravate Latency

Uncertainty
If we observe traffic at different time scales, we can see different spikes in traffic
bursts. In Figure 2-3, the 1 second-level traffic curve is a relatively flat red line
with occasional small bursts. The 100-millisecond-level traffic curve is a blue line
with slight jitters throughout and occasional severe jitters. The 1 millisecond-
level traffic curve has severe jitters from beginning to end. The rate of the 1
millisecond-level traffic can surge to over five times that of the 1 second-level
traffic at one moment and drop to zero just one millisecond later.

5
Background of DIP Networking
 Figure 2-3 Traffic bursts at different time scales on an IP network interface

The reasons for the preceding issues are as follows:

1. An IP network carries various types of services whose packets are typically

sent at irregular intervals. As a result, there is a chance that they collide on
an outbound interface. The more severe the collision, the higher the latency.
2. For some services, packets are sent sporadically (meaning that no packets
are sent occasionally) or packets are sent with severe bursts. Once multiple
services of this type collide with each other, the latency becomes especially
high.

2.2 Deterministic Transport Required

by the Industrial Internet
The industrial Internet includes an external network and an internal network.
The internal network is further divided into an Information Technology (IT)

6
Background of DIP Networking
network and an Operational Technology (OT) network, of which the latter is
further classified into a field network and a factory/workshop network. As shown
in Figure 2-4, a field network is responsible for the connection between the
Programmable Logical Controller (PLC), Input/Output (I/O) subcard, sensor, and
actuator, whereas a factory/workshop network is responsible for the connection
between the Supervisory Control and Data Acquisition (SCADA) system and PLCs.

Figure 2-4 Structure of the internal network of the industrial Internet

7
Background of DIP Networking
 An I/O subcard connects a sensor and an actuator through an industrial bus,
such as Process Field Bus (PROFIBUS) and Control & Communication Link
(CC-LINK). The type of the industrial bus used may vary between field
networks.
 A PLC is connected to an I/O subcard typically over an industrial Ethernet
protocol, such as Process Field Network (PROFINET), Ethernet Control
Automation Technology (EtherCAT), or EtherNet/Industrial Protocol
(EtherNet/IP). The type of the industrial Ethernet used may vary between
field networks.
 A SCADA system is connected to PLCs typically over a standard Ethernet and
occasionally over an industrial Ethernet.

Today's OT networks have the following problems, hindering the further

development of industrial production:

1. PLCs from different vendors open different northbound interfaces and to

different degrees, hindering the upward flow of massive field industrial data.
Currently, only about 10% of data can be sent to the SCADA system.
2. Physical PLCs are upgraded slowly, and are therefore unable to adapt to the
increasing trend of customized products.

Given this, one important approach for OT network upgrade is PLC

cloudification. However, the communication between the PLC and I/O subcard
requires extremely low latency and jitter, which is difficult to achieve with PLC
cloudification.

From the perspective of industrial control protocols, PROFINET classifies

communication into three levels: Non Real Time (NRT) communication, Real
Time (RT) communication, and Isochronous Real Time (IRT) communication, as
shown in Figure 2-5. NRT communication does not have special requirements on
jitter. RT communication typically requires a cycle time (interval at which the PLC
sends packets to the I/O subcard) of less than 10 ms, unidirectional network
forwarding latency of less than 1 ms, and jitter of less than 15% of the latency.
IRT communication typically requires a cycle time of less than 1 ms and jitter of
less than 1 μs.

8
Background of DIP Networking
 Figure 2-5 Classification of PROFINET communication types

From the perspective of industrial control application scenarios, different services

impose different latency requirements. For example, material transportation
requires a cycle time to be at the 100 ms level; machine tool control requires a
cycle time to be at the 10 ms level and the jitter to be less than 100 µs; and
some high-performance synchronization processing requires the cycle time to be
at the 1 ms level and the jitter to be less than 1 μs. The latency and jitter
requirements of these service scenarios roughly correspond to the three
communication levels defined by PROFINET.

Existing IP networks cannot meet such differentiated latency and jitter

requirements, hindering cloudified PLC reconstruction. Such requirements can
only be met with new technologies that can provide deterministic transport.

9
Background of DIP Networking
2.3 Deterministic Transport Required
by 5G to Enable Numerous Industries
In June 2015, the International Telecommunication Union-Radiocommunication
Sector (ITU-R) 5G working group determined the following three major
application scenarios of 5G at its 22nd meeting:

 Enhanced Mobile Broadband (eMBB):

 Ultra-Reliable and Low-Latency Communication (URLLC)
 Massive Machine Type Communications (mMTC)

As per the URLLC KPIs defined in 3GPP TS 22.261, services in some industries
have high requirements for low latency, low jitter, and high reliability. For
details, see Table 2-1.

Table 2-1 KPIs in different 5G URLLC scenarios

Scenario Maximum Survival Communication Reliability

Allowed Time Service
E2E Latency Availability

Discrete automation 10 ms 0 ms 99.99% 99.99%

Process automation – 60 ms 100 ms 99.9999% 99.999%

remote control

Process automation – 60 ms 100 ms 99.9% 99.9%

monitoring

Electricity distribution 40 ms 25 ms 99.9% 99.9%

– medium voltage

Electricity distribution 5 ms 10 ms 99.9999% 99.999%

– high voltage

Intelligent transport 30 ms 100 ms 99.9999% 99.999%

systems –
infrastructure backhaul

10
Background of DIP Networking
In addition, TS 22.104, TS 22.186, and TS 22.289 defined KPIs for the application
scenarios of network physical control, Internet of Vehicles, and railway
communications. Many of these scenarios require a maximum allowed E2E
latency of 10 ms.

It is difficult for existing IP networks to meet the requirements of the preceding

scenarios in which the latency must be less than or equal to 10 ms. To this end,
new technologies are required to provide deterministic transport.

2.4 Emergence of DIP Networking

To provide deterministic transport on existing IP networks and meet the
deterministic transport requirements of the industrial Internet and 5G vertical
industries, Huawei proposes DIP networking.

11
Background of DIP Networking
 Chapter 3
Value of DIP Networking

Abstract
DIP networking helps facilitate the continuous evolution of the
industrial Internet, advance the access of all industries to 5G networks,
and accelerate new service incubation.

3.1 Facilitating Evolution of the

Industrial Internet
As shown in Figure 3-1, a DIP network is used for the communication between
the cloudified PLC and I/O subcards, meeting the latency and jitter requirements
of production and control services on the OT network and facilitating the
evolution of the industrial Internet's OT network.

12
Value of DIP Networking
 Figure 3-1 Evolution of the Industrial Internet's OT network structure

The cloudified PLC system provides the following benefits:

1. A large amount of field industrial data is obtained and stored on the cloud
through the cloudified PLC system. This facilitates data transmission and
exchange with the SCADA system and even the systems on the IT network,
and subsequently enables industrial big data analysis.
2. The transformation from the physical PLC to IT-based software system is
implemented, which accelerates PLC upgrade and better adapts to the
increasing trend of customized products.

13
Value of DIP Networking
3.2 Advancing All-Industry Access to
5G Networks
The rapid development of 5G has led to fast-tracked digitalization of various
industries, including electric power, transportation, finance, education,
healthcare, mining, and ports. Each industry has a wide range of service
scenarios, which have different requirements on deterministic forwarding
latency. The strict deterministic latency that some of them require on the
transport network makes traditional IP forwarding technologies unviable.

Figure 3-2 Differential power protection

In the differential power protection scenario shown in Figure 3-2, protection

relays exchange real-time current data, and each relay compares the received
data with its local current data at the time point when the peer relay sends the
data to detect whether a fault occurs. This type of data communication has strict
requirements on the latency (less than 5 ms) and jitter (less than 200 µs). These
requirements cannot be met using traditional IP networks. In light of this, DIP
networks are required.

Such networks provide deterministic forwarding capabilities and support network

access for various industry scenarios that have strict requirements on latency and
jitter. The specific benefits are as follows:

1. New services, such as remote driving, remote mining, and remote medical
treatment, can be deployed no matter the distance, which in turn maximizes
resource utilization and allocation.
2. Existing IT technologies, such as big data analysis and intelligent analysis,
can be utilized to reconstruct existing industry services, improving efficiency.

14
Value of DIP Networking
3.3 Accelerating New Service
Incubation
When the Internet was just established, it would have been hard to imagine that
PCs, mobile phones, watches, automobiles, home appliances, and more could
access the Internet.

Likewise, when mobile phones were first launched, it would have been hard to
imagine that they could offer so many applications to make life easier, such as
shopping, travel, video, financial management, and online courses.

This is where we are now with DIP networking. It is still in its infancy, but its
potential to provide deterministic forwarding capabilities with bounded jitter and
bounded latency has already been demonstrated. It has been proven to meet the
microsecond-level latency and jitter assurance requirements of a variety of
future-oriented services, such as industrial automation, industrial teleoperation,
holographic communication, and vehicle-road synergy. Building on this, it will
play a big role in promoting the smart manufacturing strategy, accelerating
industry digitalization, and ultimately improving economic competitiveness.
Given time, this promising "fertile soil" will gradually grow into a lush,
blossoming garden.

15
Value of DIP Networking
 Chapter 4
Overall Architecture of DIP
Networking

As shown in Figure 4-1, the overall DIP networking architecture consists of three
parts: edge shaping, interval mapping, and SRv6 explicit path planning. Edge
shaping and interval mapping are used on the forwarding plane, whereas SRv6
explicit path planning is used on the control plane.

16
Overall Architecture of DIP Networking
 Figure 4-1 Overall architecture of DIP networking

 Edge shaping shapes traffic on the ingress Provider Edges (PEs) of a network.
Through edge shaping, packets with irregular arrival times are shaped to
different T intervals. Edge shaping eliminates the dependence on strict
packet entrance time. It allows packets to enter a device at irregular times
rather than only at strict intervals. It also supports burst traffic.
 Interval mapping is responsible for controlling the latency of packet
forwarding hop by hop on the Provider (P) devices or egress PEs on a
network. Through interval mapping, the packets sent from an upstream
device within an interval of T are sent out within another interval T by the
local device's outbound interface. Interval mapping allows a high
transmission latency on links between devices and therefore supports large-

17
Overall Architecture of DIP Networking
 scale networks. In addition, interval mapping only cares about the interval
— not the flow — to which the packets belong. This simplifies the
implementation on the P devices, making it easier to expand their flow
specifications and simplifying the identification of inner flows by the P
devices when various types of tunnel encapsulation are performed on a
Layer 3 network.
 SRv6 explicit path planning is used for controlling forwarding paths and
reserving hop-by-hop forwarding resources. This is essential for achieving
deterministic latency.

As shown in Figure 4-2, a packet is shaped to interval T0 on the ingress PE, and a
jitter interval of T is generated. Through hop-by-hop interval mapping with a
fixed offset across multiple transit P nodes, the latency is always deterministic.
On the egress PE, the packet is sent within another interval T. The packet is the
first one in a specific interval on the ingress PE, and becomes the last one in
another interval when it reaches the egress PE due to the packets sent from
other P nodes being placed in front. In this case, a jitter interval of T is
generated. DIP networking can always control the jitter during E2E packet
forwarding within two T intervals, demonstrating its accurate jitter control
capability.

18
Overall Architecture of DIP Networking
Figure 4-2 Jitter implementation of DIP networking

19
Overall Architecture of DIP Networking
 Chapter 5
Key Technologies of DIP
Networking

Abstract
This chapter describes the key technologies of DIP networking. Among
them, gate scheduling is the foundation; edge shaping, interval
mapping, and SRv6 explicit path planning are used together to provide
deterministic capabilities; and dual fed and selective receiving is used to
improve network-level reliability.

5.1 Edge Shaping

Edge shaping is used to shape multiple packets of a flow into a suitable gate
control queue. These packets can be sent only when their gate control queues
are open. In Figure 5-1, four packets belonging to one flow arrive within a short
period, but each interval reserves only three positions for the flow packets. At
the current time, the gate that can first accept packets is gate 3. In this case,
packets 1, 2, and 3 enter gate 3, and packet 4 enters gate 4.

20
Key Technologies of DIP Networking
 Figure 5-1 Edge shaping

5.2 Gate Scheduling

As shown in Figure 5-2, gate scheduling provides eight gates, as well as one gate
control list that controls the open/closed state of the gates as scheduled.

21
Key Technologies of DIP Networking
 Figure 5-2 Gate scheduling

 A DIP flow uses six gates, which are opened one by one at an interval of 10
μs. After reaching gate 5, it cycles back to gate 0 for the next 10 μs. This
process repeats.
 A common IP flow uses two gates, which are always in the open state.
 Strict Priority (SP) scheduling is configured for the eight gates. When the
gate control queues for a DIP flow are opened, DIP packets (if any) are sent
first.

5.3 Interval Mapping

As shown in Figure 5-3, after edge shaping is performed on the packets on the
ingress PE, they are sent through the outbound interface within interval T0. After
arriving at P1, the three packets in interval T0 are sent through P1's outbound
interface within interval T2. Similar interval mapping and forwarding are
performed on each hop until the packets are sent out from the egress PE.

22
Key Technologies of DIP Networking
 Figure 5-3 Interval mapping

5.4 SRv6 Explicit Path Planning

Gate scheduling, edge shaping, and interval mapping lay a foundation for E2E
deterministic latency of packets passing through the network. But without
deterministic paths, E2E deterministic latency cannot be achieved. To provide
deterministic paths, SRv6 explicit paths need to be planned.

23
Key Technologies of DIP Networking
 Figure 5-4 SRv6 explicit path planning

SRv6 explicit path planning includes the following steps.

1. Network Cloud Engine (NCE) obtains device information, device connection

modes, and interfaces and bandwidths that support DIP forwarding.
2. A network access flow is configured on NCE, with requirements on the flow's
network ingress, egress, rate, burst size, and jitter specified.
3. NCE calculates whether an eligible path exists on the network. On this path,
the remaining bandwidth reserved for DIP forwarding on each device's
outbound interface that passes through the path must be sufficient, and the
E2E latency and jitter of the path must also meet the requirements. If such a
path exists, the flow is allowed to access the network and occupy the
corresponding resources.
4. NCE delivers flow identification, traffic shaping, and SRv6 explicit path
information to the network ingress.

24
Key Technologies of DIP Networking
5. Packets of this flow enter the network and are forwarded along the planned
explicit path. The E2E forwarding latency and jitter can then meet the
requirements of this flow.

5.5 Dual Fed and Selective Receiving

Dual fed and selective receiving means that a packet is replicated into two copies
on the ingress PE and then forwarded along two paths. The egress PE selectively
receives one copy for packet forwarding.

The network is bound to have a certain failure rate. On the network shown in
Figure 5-5, traffic passes through the ingress PE, P1, and P2 to reach the egress
PE. The traffic traverses three links in total: ingress PE-P1, P1-P2, and P2-egress
PE. If we assume that the reliability of each link is 0.99999, then the reliability of
the E2E link that the traffic traverses is the product of the reliability of the three
links: 0.99999 x 0.99999 x 0.99999 ≈ 0.99997. When high traffic reliability is
required, dual fed and selective receiving can be used to improve the overall
reliability.

Figure 5-5 Single-path link fault

On the network shown in Figure 5-6, a traffic flow arrives at the egress PE over
two paths. The path ingress PE-P1-P2-egress PE is path 1, and its reliability is
0.99997. The path ingress PE-P3-P4-egress PE is path 2, and its reliability is also

25
Key Technologies of DIP Networking
0.99997. The reliability calculation method of path 2 is the same as that of path
1. In dual fed and selective receiving mode, E2E traffic forwarding is normal so
long as one of the two paths is normal. Therefore, the formula for calculating
the reliability of dual fed and selective receiving is as follows: 1 – (1 –
0.99997) x (1 – 0.99997) = 1 – 0.00003 x 0.00003 = 1 – 0.0000000009 =
0.9999999991. We can see from the above that dual fed and selective receiving
greatly improves the reliability of traffic forwarding.

Figure 5-6 Dual-path link fault

5.6 Summary
A DIP network uses edge shaping, gate scheduling, interval mapping, SRv6
explicit path planning, and dual fed and selective receiving technologies to
ensure that the jitter is bounded and small, the latency is bounded, and the
reliability is very high — regardless of when a packet flow reaches the network
ingress and how many hops it traverses before being forwarded by the network
egress.

In industrial Internet and 5G vertical industry scenarios, DIP networks well meet
the requirements of jitter- and latency-sensitive services. The bounded jitter and
latency of a DIP network can be calculated theoretically and guaranteed in real-
world scenarios, ensuring What You See Is What You Get (WYSIWYG).

26
Key Technologies of DIP Networking
 Chapter 6
Applications of DIP
Networking

Abstract
DIP networking has been tested in China Environment for Network
Innovations (CENI) and China Telecom Shanghai's cloudified PLC
project, where it has shown excellent results. These results have been
released at the Future Network Development Conference and the
Global Network Technology Conference, attracting wide attention from
the industry.

6.1 CENI
Huawei and Nanjing's Purple Mountain Laboratories (PML) are long-established
partners and co-conducted DIP networking tests on the CENI. As shown in Figure
6-1, the test traffic first traveled from the PML in Nanjing to the CENI lab in
Beijing through numerous network nodes, including Hefei, Wuhan, Zhengzhou,
and Shijiazhuang, before traveling back to the PML through the same network

27
Applications of DIP Networking
nodes in the reverse order, spanning a total length of 2000 km. Bidirectional
interference flows were injected using testers at the two sites.

Figure 6-1 CENI networking test

Figure 6-2 shows the test results: The E2E jitter is large on a traditional IP
network but small on a DIP network.

28
Applications of DIP Networking
 Figure 6-2 Latency of the CENI loopback test between Beijing and Nanjing

When different numbers of interference flows were added, as shown in Figure 6-

3, the maximum jitter increased as the number of interference flows increased
on the traditional IP network but remained stable on the DIP network. This
indicates that jitter performance on a DIP network is not related to the number
of interference flows.

29
Applications of DIP Networking
 Figure 6-3 Jitter comparison of the CENI loopback test between Beijing and Nanjing

On August 15, 2020, the PML released the world's first test results on
deterministic WAN innovation at the 4th Future Network Development
Conference. The PML stated an E2E experimental jitter of less than 30
microseconds was achieved for 2000 km network transmission through 13
network nodes between Nanjing and Beijing. This was a world first. They also
noted that this technology lays a solid foundation for implementing super-long-
distance control. It's a milestone in industrial Internet development, and will
pave the way for the Internet to achieve even greater success in the real
economy than in the consumer field.

6.2 Cloudified PLC Project of China

Telecom Shanghai
Huawei cooperated with China Telecom Shanghai to test DIP networking
performance in a cloudified PLC project for a super large enterprise. In this
project, there were four hops between the control and actuator ends, and the
transmission distance was 22 km.

30
Applications of DIP Networking
 Figure 6-4 Networking test for China Telecom Shanghai's cloudified PLC project

As shown in Figure 6-5, the test results demonstrate that for traditional IP
forwarding, the latency is low when the load is light and is high when the load is
heavy. However, for DIP networking, the latency is always low regardless of
whether the load is light or heavy. Further to this, as shown in Figure 6-6, when
the network is heavily loaded, the jitter of traditional IP forwarding is large, with
obvious fluctuations. However, for DIP networking, the latency is stable, with
little jitter.

31
Applications of DIP Networking
Figure 6-5 Latency comparison between traditional IP and DIP forwarding in light and heavy
load scenarios

32
Applications of DIP Networking
 Figure 6-6 Latency comparison between traditional IP and DIP forwarding in heavy load
scenarios

On December 23, 2020, China Telecom Shanghai delivered a speech titled

Deterministic IP Network Oriented to the Industrial Internet at the Global
Network Technology Conference. The speech stated that no matter whether the
load was light or heavy, there was almost no jitter for DIP. Traditional IP had
obvious jitter under the same background traffic, which proved that the latency
of DIP forwarding was not related to the link load.

33
Applications of DIP Networking
 Chapter 7
Future Prospects of DIP
Networking

Abstract
Even the largest trees start as a seed.
There are many factors that determine whether a technological concept
takes shape. Broadly speaking, national policies and industry standards
are required. Narrowly speaking, key technology breakthroughs are
needed. More importantly, there must be a consensus around the
concept. The same is true for DIP networking.

National Policies
Over the past two years, China has released several influential documents on
accelerating digital industrialization and promoting economic and social digital
transformation.

These policies encourage industrial enterprises to explore new technologies, such

as cloud-network synergy, deterministic networking, and SRv6, with the aim of

34
Future Prospects of DIP Networking
reconstructing enterprises' intranets and promoting the convergence of IT and
OT networks.

Industry digitalization and IT-OT convergence of the industrial Internet will

generate many new requirements on networks. Exploring new DIP technologies
is a new direction favored by national policy makers.

Industry Standards
Over the past two years, China has approved the DIP networking standards
initiation work relating to the overall architecture and technical requirements for
DIP networking of telco networks as well as DIP technical requirements on telco
network devices.

The approval of standard initiation signifies that Chinese carriers and

communications companies in the industry widely recognize DIP networking,
greatly boosting the prospects of DIP networking.

Breakthroughs in Key Technologies

DIP networking has witnessed significant breakthroughs in its overall
architecture and technical implementation. Tests in some industries have already
verified that the technologies can meet the requirements of new services. The
next move for DIP networking is commercial deployment to support the
operation of new services before gradually expanding the application to the
networks of different industries and customers.

Consensus Formation
DIP is supported by national policies, related industry standards are being
formulated, and breakthroughs in key technologies are being made. As such, we
believe that with the propagation of related technical knowledge, a consensus
on DIP networking will eventually be reached among all stakeholders in the
industry to promote its comprehensive application.

35
Future Prospects of DIP Networking
Contact Us
networkinfo@huawei.com

More IP Network eBooks

https://e.huawei.com/en/solutions/enterprise-networks/ip-ebook

36
TOF/TOT1