 Good morning and good evening. Welcome to ONIF, as we call it, Open Networking and Edge Executive Forum. We are really excited. This is day two of the conference. As you know, we have set this up in three different time zones so that people within that time zone can attend and listen to some of the top executives who are speaking and on behalf of ONIF. So thank you very much for joining us. I'm your host, Arpit Jasper. I head up LF Networking and LF Edge, among other projects at the Linux Foundation. And we have a very exciting lineup of speakers today. They are top leaders from Asia. And they are all very well known leaders in the telecommunication edge and the cloud space. So we're going to cover a lot of ground today. And without any more delay, I want to introduce our first speaker. But before I do that, a few housekeeping notes. These are live. So if you have questions, please ask them through the chat. And it's located. The Q&A button is on the top right of your screen. Also, this is being broadcast in 16 languages. So if you want to hear in a different language, there's a button below the world widget. So if you look at the bottom, there's something on your screen called world widget. Click there and pick your language. And if there's any questions, there is always the events staff, LF events staff that will be online to answer any FAQs. So with that, let me introduce one of the first speakers today. We have four in this session. We will have a break. And then we will have three more very, very powerful lineup. The first speaker is Mark Shen, who is a senior technical manager at Tencent and obviously a board chair for TARS Foundation. Tencent, as you know, does not need any introduction. They are very active in our edge community and our cloud community along with the TARS Foundation. So without any further delay, please welcome Mark Shen. Okay. Good morning, everyone. I'm Mark from Tencent. I'm the chair of Tencent Open Source Alliance and the chair of TARS Foundation, Gavin Bo, and also the TSC member of Lina Foundation, Aquareno Aztec. It's my pleasure to share with you today about Tencent's practice on cloud and edge. Okay. Let's begin. Okay. First of all, let's discuss why do we need edge computing. When it comes to cloud computing, everyone will immediately bring up central cloud computing. Central cloud computing is a central architecture and the computing resources are located in the central computer room and maintained by the cloud vendors. On the other hand, edge computing is a kind of disputed computing. Computing resources are scattered closer to the data source to achieve the purpose of providing services nearby, mainly because the needs of scenarios are constantly changing. Many traditional industries have put forward more new information, transformation needs. In general, edge computing can benefit for aspects like larger capacity, lower latency, lower cost, and support for localized processing. And it will be widely used in different areas like cloud gaming and smart office, industry, internet, and so on. But there are also some new challenges that come with the edge computing scenarios like heterogeneity is substantial. This is reflected in both software and hardware. For example, the central cloud and edge cloud usually use X, AES, and linear standard releases. In contrast, edge resources are likely to be cheaper or customized software and hardware due to the need considered cost and special business requirements, as you know. And the second is that the scale is huge. According to organization, the number of global IoT devices will exceed 100 billion in 2025. This will be worldwide. How to manage that last scale equipment is also a very challenging task. And the third one is that the environment is complicated. The equipment located in the cloud computer alone is okay. And a large amount of terminal equipment is often located in harsh environments. For example, a lot of equipment in steel mules is in high temperature environments for a long time. The device network environment is also various weird and wireless like Wi-Fi, 4G, and 5G network, ZB, and etc. The first point is that the standards are not unified. There are no standards in many places or there are many standards, but not widely you can recognize standard, especially in management methods. The rest of all these challenges are like you can see in the slide, low efficiency, difficult to manage, and low reliability. Now I will take about how Tencent faces these challenges. Let's have a look at Tencent cloud and edge infrastructure. The central cloud usually refers to the IDC computer and the edge cloud will be the EC, OC, Mac, computer in turn. Few devices are generally located near the data source, such as home-gameway traffic light interaction. Generally, the delay between the IoT device and the edge device can be controlled within two milliseconds. It is suitable for processing business data with extremely high real-time requirements, such as industrial control services. The delay with the edge cloud can be controlled within 10 milliseconds, which can meet the requirements of real-time audio and video business scenarios, AR, VR, and cloud gaming. Tencent has accomplished many use gains based on this infrastructure, such as Tencent's cloud gaming products, like regain, start cloud gaming, and the HD video, and live streaming, like Tencent video, and we have a product called VC, the connected car, and the Italian navigation, and Tencent meeting, and live zoom, and we have Tencent documents for the smart office, and Tencent IoT link is for the IoT. Besides, Tencent also helps a lot of products from other companies that partner with Tencent. They are in different fields, like education, live streaming, and marketing, and I will show you later. Tencent was hard to manage those challenges through hardware and software ways. About the hardware, Tencent has Edge Computing Machine ECM, which provides low latency, high performance, reliable computing, and the network cloud services, and large nodes close to the user's location. Also, we have the Edge Computing Center, that is the mini-T-Broad, which is lightweight, deployment-providing comprehensive innovation, and deliverable 5G Edge computing over all solutions, and you know, T-Broad uses many in China and in our industry also. And we also have the Edge Intelligence Gateway ILT-Bainway is a lightweight Edge computing platform with multiple assess methods and support cloud-related services. Let's take a closer look at the hardware, as I mentioned just now. Edge Computing Machine ECM is a product that provides low latency, high availability, and low cost Edge computing services by syncing computing power from the central node to the edge nodes close to the user's. Caught in more than 300 nodes have been constructed, covering the whole region of China. And we have the Edge Computing Center uses the Tencent CloudStuff developed mini-T-Broad, a mobile data center infrastructure as a carrier, 5G Edge Computing, and the Internet experience. And it also introduces Tencent Cloud Edge computing like the i-SaaS platform. All this product capability to support the cloud 5G to see to the customer or to the business services of the games, the 4K live broadcast, and also the robotics and providing comprehensive innovation and deliverable 5G Edge computing over all solutions. The next one is the ILT-Gainway. This product is Tencent's industrial-grade improvement-facing ILT Edge application scenarios. It provides ILT device-assessed AI local enterprises and Edge cloud collaboration and other functions. It also has a small size, high reliability, and multiple networks, actual cloud and easy management. In terms of location scenarios, it is suitable for park security and also smart retail, like the power inspection and smart street lights, speed transportation, and also the water conservancy monitoring, and also like the industrial quality inspection and other scenarios. Like about the software Tencent open source, several projects like the SuperEdge and TAS. SuperEdge is an open source container management system for Edge computing to manage computer resources and container application in multiple Edge regions. Open source three months ago and with all our members that you can see in the slide. And about the TAS, that is a high-performance microsystem framework based on main service and TAS protocol, with multiple programming languages, and also an integrated administration platform with implemented hosting services. We have flexible schedule. TAS was open source in 2013, and also we found a foundation next year. And yesterday, that is the first annual story of TAS Foundation, and we have a good celebration for TAS. And TAS supports like ARM, XACs, and multiple platforms, including Mac OS, also Lightningness, and Windows, of course. And let's look at SuperEdge. The SuperEdge has the following characteristics, like the Kubernetes network. Of course, SuperEdge extends the powerful container orchestration and scheduling capabilities of Kubernetes to the Edge. And the Kubernetes users can leverage SuperEdge easily for Edge environments with minimal learning. And like the second point is that as autonomic and when the network connection between the Edge and the cloud is unstable or the Edge node is offline, the node can still work independently. And the third function is that like disability, no healthy monetary, SuperEdge can continue to monitor the process on Edge slide, on the Edge side, and they collect healthy information for faster and more accurate, popular discovery and reporting. And it can also build in Edge orchestration capability, SuperEdge supports automatic deployment and multi-region microservices. Edge side services are closed looped and it effectively reduces the operational overhead and improves the fault currents and the availability of the system. And the last one is the next Turing. SuperEdge ensures that Kubernetes nodes can operate under different network situations. It supports the network Turing using TCP, HTTP and HTTPS. The second open-source project is TAS. TAS can quickly build systems and automatically generate code taking into account easily of used high performance. At the same time, TAS supports multiple program languages, like you can see in the slide, including C++, Golang, Java, Node.js, PHP and Python. TAS can help developers and enterprises quickly build their own stable and reliable disability application in microservices manner so that they can focus on business logic to improve operational efficiency effectively. And TAS is working in TAS level. It can run on physical machine, virtual machines, and also containers and including the Docker and Kubernetes. You can store TAS data in cache, also like DB and the other devices. TAS framework supports TAS protocol and also like the other protocol QP, SSL and HTTP one or two, like the PB protocol, buffers and other customized protocols. TAS protocol is an IDL-based binary and it's tangible and cost platform. This feature allows servers to use different languages to communicate with each other using RPC and TAS supports different RPC methods like sync and one-way request for these methods. TAS has multiple functions to government services and can integrate others such as like the Jenkins and Python and other open-source projects. Additionally, Tensen plays different roles in communities that support edge computing such as Acreno. And as most of us know, Acreno is a set of open infrastructure and application brokings for edge. There are a variety of use cases, including 5G, AI edge, TAS, IoT for both providers and enterprise edge domains. The acreno community has worked together to provide resources to developers and open source participants to ease the development across the different hardware platforms and architectures. Tensen has contributed to several brokings in the Acreno community. They are the connective vehicle brokings which focuses on establishing an open source MAC platform. The backbone for V2X application, the implement functions like accurate location, smarter navigation, and safe drive improvement and reduce the traffic violation and the exact chart. We have also the other ARVR brokings and this proofing simulates Acreno to improve the teachers and students online education experience. The 5G MAC and slide system brokings is a target to support cloud gaming, HD video and live broadcasting. And let's see about the case and some use cases in Tensen's practice. The first one is the cloud gaming. Edge computing machines can improve the user experience by deploying cloud game servers on edge nodes, reduce the delay of use operation feedback and improve the fluency of the picture. And the end-to-end interaction of the game is strong. The edge of ability is connected to the nearest public network through the local operator line. The server in the main node is assessed through the Tensen cloud private network. And like reducing the latency and enhancing the user experience at the same time, it can achieve multiple point disaster recovery on the public network's assessed side and reduce the bandwidth cost of the public network. The second one is the video live streaming. It is from Huya, the Tensen's partner with Tensen cloud edge computing nodes covering the whole region of China in relying on high-performance computing and low-latency network products and services. Huya provides a low-latency high-definition stable interaction video and live streaming experience for large number of users. During the live broadcast edge computing transcoding and destruction have been implemented, Huya provides main services and low balance with support automatically during after-definition. And while obtaining a low network delay, the center's pressure is reduced and the central bandwidth cost is saved by more than 30%. And the network connection delay of edge nodes is less than five minutes, which provides the uplink quality and viewing experience. Okay. The last case is that smart CPU water sales. And here I take intelligent water pollution fashion as the use case. The system can dynamically model and analyze the reading of the computing image, organization and the vision technology. The scope of the supervised water area is obtained by the skin semantic segmentation or user defined way. And the system realize intelligent water management form of a normal sewage disaster, sewage exchange and reward and water source supervision by detecting and classifying the disaster post with your picture. We can judge whether there is a sewage exchange behavior. With the taggy detection algorithm, we can get the real-time location or throwing objects or people or ship or ship among the coast. In the water area, we can analyze whether there is an injection or other. Or create the early warning and realize the efficient and dynamic management of water area supervision. Okay. That's all. Yeah, thank you. Thank you. Thank you, Mark. That is very insightful. A lot of the blueprints in Acreno are very exciting and Tencent is really, really leading this. So thank you very much. We're going to move forward now with the next presentation. Bill Ren, who is the chief open source liaison officer at Huawei GM in the ICT infrastructure group. A very, very strong leader he's been presenting and his vision and the leadership that they provide is extremely good. So without any further delay, Bill, go ahead. And all our friends. And good morning to everyone. I'm Bill Ren and from Huawei. I'm glad to share our thoughts about how to enrich the 5G to be business and make sure the industry knowledge system can bridge the gap of digitalization. Okay. Please go to the next page. And as you can see, we all know the fourth industry revolution is arising. And but we do need to realize this time the fourth time industry position is totally different as the first three runs, no matter its age of steam or electricity or information. Before it's all from supply side. It's focused on more productivity demands, like the productivity quality, efficiency and the consistent. But this time, it's more driving from demand side. It's more focused on value. And then the information personalized requirement and diversity requirement is more important. So this require a platform to do this. So next page, I will talk about how to do this. As you can see, the 5G network is constructing globally and the collaboration of connection and computing is key for enabling the industry to do the digitalization. So that's why we think 5G should serve as a cornerstone for the first industry revolution. And as well, in the middle, we think a unified and open MEC platform is crucial to make sure we can promote those industry applications to go a large-scale application, to go mature. So that's why we believe the 5G MEC solution is very critical. And we do think the unified platform is important. So next page, but what's the challenge I will talk. Can you go to the next page please? Yeah. As you can see, 5G2B has do a lot of a proper concept project globally. It's claimed that there are more than 6,000 POCs ongoing in the past two years. They are focused on so many scenarios from call manners, HD video, and a lot of smart campus, hardware, and so on. But still, people think that it's difficult to do the replication for those applications in a large scale. So what's the reason? What's the challenge we are facing? We believe because there is no unified platform. So all the innovation, we start and do it from scratch every year. And there is no closed loop of business. So the motivation for doing those industrialization and revolution is to make people more frustrated. So that's why we think a unified platform is needed. So next page, I will talk about what's the key reason behind this. You can see, we clearly see during those projects, the huge gap between the industry knowledge and the 5G, ICT technologies. You know, in 5G or in ICT technology, we all talk about 5G cloud, machine learning, and others such as superlink, the dynamic loading. But for the industry, they could not understand those technologies. They are experienced, expertise, more focused on industry know-how, such as the project product control, quality control, and many other industry process and sourcing related scenarios. So just like what's the famous point, the point is to give this big, famous thing. The first distance in the world is not when I stand in front of you, you cannot see me or understand what we are talking. So that's the case, what we are doing. It's happening in the ICT industry. So next page. But how to solve this problem? I will talk about it in the next page. So you can see, building this industry knowledge ecosystem is very critical. So the industry knowledge experience and expertise is just like this pyramid from ERP to MES to SCADA to the sensor controlling and even some field manufacturing. But how to replicate those knowledge and experience? You even use success in one project because it's so huge, it's difficult in such large granularity to replicate it. So we have to decomposite it to make sure we achieve a lot of atomic capability, just like in the software microservice-based component library. So this could make the project and the knowledge very reusable in large scale. So you can see we can, based on this microservice-based component, we can reorganize it based on the new scenarios, based on the new process to do this precipitation. Then we can achieve those applications in a more large scale, make it more possible and feasible. So that's why we believe we do need to build such kind of capability in the whole industry. So next page, we will see how to do it. Again, another observation we found during the first industry revolution, digitalization. We found we should not only do the digitalization in the past, we like project based. So the digitalization should be happening in cross industry collaboration. So the cross industry collaboration, the most important is the platform interconnection. Because every industry, every vertical industry, especially those top players, they all build their platform. Some of them even aggregated in industry level. They have some kind of unified platform, like Predix, like Siemens, like some other Bosch, they all have a different platform. So they all want to do the digitalization and reuse their expertise and experience. So how to make sure this platform, this expertise can leverage the ICT technology. So we need this knowledge, software precipitation, with those API capability from ICT point of view, based on the new business scenarios. So there is a lack of an industry platform to bridge this gap for the ICT conventional platform. So next page, you can see in the left hand side, the industry requirement based on different vertical industry knowledge. And the right hand side, the technology supply based on the open source practice. So we believe open source and openings will be more effective methods for building such a unified platform to make sure the OT, IT plus the CT can work together very well. Okay, next page. That's why we think in the 5G MEC area, that's why we build this open source platform. We call it edge gallery open source project. It's our 3K value. First is the 5G native. Make sure this 5G native connection capability and the 5G more accessible. And the second is edge native platform architecture. Make sure those services trustworthy. And the third one is the diversity to open edge ecosystem. So we build this architecture from design time to runtime. Make sure it's a closed loop based on the edge security and edge infrastructure in place. So here, I'd like to show a video to quickly give you a view about how this edge gallery looks like. Please play this video. Thank you. Edge computing empowers an intelligent society by becoming great avenue for carriers, enterprises and vendors to build new opportunities via best utilization of 5G. Currently, 5G MEC application development, integration and commissioning poses a lot of difficulties due to heterogeneous platforms. Open unified 5G MEC platform is need of the hour and edge gallery is solution in this direction. The edge gallery platform community focuses on the 5G edge computing scenario through open source collaboration to build the framework of resources, applications, security, management of the multiple edge locations, de facto standard for network open services to achieve interconnection with public clouds and build unified MEC application ecosystem based on the heterogeneous compatible edge infrastructure. To achieve this edge gallery constitutes four major modules each focuses on a set of functionalities pertaining to the MEC application development and deployment. Multi-access edge platform, autonomous, that is rich in platform and network capabilities, support heterogeneous architecture. MEC manager, helps in application orchestration, application lifecycle management, F caps and policy driven closed loop actions. MEC developer platform, provides tool chain, SDKs, APIS and sandbox features for application development, packaging and testing. MEC application repository, for unified app package format, standardized API between operator app store and MEC manager. Edge gallery is one stop platform catering to various types of users by supporting quick integration of existing applications, integration of operators application repository to build federated repository and rapid development of new applications on both x86 and ARM architecture. The developer platform contains a series of developer tools and plugins and provides abundant 5G telecom network capabilities for developers at the edge. The edge gallery platform supports commissioning of applications on both the simulator sandbox with simulated MEP and real 5G lab environment which includes the actual 5G network and UPF to allow users to test applications in real 5G edge scenarios. All application are authenticated and tested for standard and security compliance by application test platform and then publish into app repository. MECM management plane takes care of edge nodes management, application package distribution to these nodes and application configuration rules. Complete edge node topology map on live network can be viewed and monitored in MECM the map clearly displays the deployment location of edge nodes, resource usage, hardware and software capabilities, deployed applications and service information on each node. To leverage the collective investments being made by various edge gallery community partners, edge gallery has introduced the lab bus service which allows the community test lab to operate with much higher efficiency as a result of their larger scale, dedicated management, and proper tooling. This provides significant capex and opets savings and also improves the quality of tests being conducted. Edge gallery community is expanding with currently 35 plus participating companies with 111 authors and more than 7500 commits. Come join us on this great journey of open source collaboration to build a world-class unified 5G MEC platform. Yeah, so next page. So thank you for watching this video. You probably maybe got a big picture of what edge gallery looks like. You can see it's more focused on application and preparation, integration, testing, and application release and the deployment. So we can build a mirrorly 5G environment in our lab in the community to help the application developers, industry players, enterprise operators to build this application repository together and then we even could deploy it in an industry production environment in Kerr's network. That's our intention. So next page, as you can see, the edge gallery so extends this 5G capability to the edge and promotes this knowledge ecosystem. And edge gallery is more focused on the multiple roles. From developer side, we focus on developer environment and this application developers and traders more focused on application repository and operator side in this preliminary environment to do this management and the deployment of MEC and MEP and also as well as this enterprise customers such as we have two use cases here, one for to be a business like as a robot company to how to deploy that robot application in our platform. The other is VR players how to deploy VR applications in our platform. So next page and edge galleries has started to formally open sourced in last day August and we were actively cooperated with the industry mainstream standards and open source organizations such as 3GPP, GSMA open platform and ETSI MEC standards as well as Linux foundation is a key player in the open source. We actively involve in the edge umbrella project and we build our blueprint in the community. So also we work with other events such as 5GDNA, ECC and 5GAI and other organizations. Okay, next page. So finally, as you can see the 5G is just started 5G2B is a beginning as we always know a journey of a thousand miles begins with a single step. We believe this first industry revolution and digitalization is a long way to go but we do need to look at it in a more growing way just like growth in the in the soil. So we have to do it in a right way to build a common open source community to to reach the common understanding, common framework, common architecture to lamp to bridge those industry ecosystem with a technical exit as well as this business exit to make sure the developers, the industry expertise and the end user for enterprise to work together in a sustainable and continuous way to make sure we can reach the target. So that's why I ask for all the industry players especially for the technical operators. The no matter is that you are from technical side you are from equipment manufacturer window side or you're from enterprise side we need to work together in an open and software way to make sure the force digitalization can be reachable and it can be feasible and we believe this is the right thing to do. Thank you. Thank you, Bill. Thank you. There's I know we we have very limited time but there are a couple of questions. I would like to ask you one quick one if you could answer it quickly. Do you think that edge gallery is kind of needs to be done vertically different in different regions of the world or so does edge gallery require localization or does it have a unified approach across the world? Sure that's a really good question. As we say open source we believe from beginning it's showed by natural it should be global. So at the beginning once we think of should do the open source community open source platform open source project we always think it should be global but as you can see from past one year those global the pandemic since we start from Asia area mainly from mainland of China we focus on top three operators in China at the first step but we already started working with our European partners and as well as other human other areas so we are definitely from our initiative target it should be global. Okay very very good. I think there was another question in a similar way in which I'll probably answer as well which is you know does this require you know does companies and effort require a partnership with US and Europe companies or you know do you do it on your own and I think the answer is very clear right Bill said it very well. Yes yes open source is a global collaboration. Yeah one more thing is as you may say we open source it in Haiti actually we always do it also in GitHub so it's a tooling could be easily reachable and yeah very very good thank you very much Bill very insightful glad to have you on board. Thank you thank you. Okay so moving on the next speaker no introduction and you know he is probably I would say single-handedly if not with his whole team turned the world upside down of telecommunications right and and you know I think from the Geo days to the Rakuten days you know he's a thought leader which we all who we all admire we look forward to hearing him all the time he's always got some you know some crazy some real some actually very cool ideas that we all like to hear and borrow so without any further delay please welcome you know Tarkham in CTO of Rakuten. Thank you very much Arpett nice to be with you today as usual I really really enjoy this type of forms because it gives me a path to discuss you know the main building blocks of how we envisioned having an open disaggregated cloud native network so with that to do let me start my presentation just giving everybody a little bit of understanding about what is really Rakuten before we are a mobile operator I think I've spent a lot of time talking about this before and one of the primary reasons that attracted me to move from India and come to Japan was really this large massive ecosystem a company that really has in the beginning fundamentally no business to do with telecom it is an internet facing company that is in banking and e-commerce in digital payments sports because we believe that has a big impact on branding and various content and media application so it is to my knowledge a first time that an OTT internet company had a desire to really enter connectivity for me personally I thought this is a great opportunity to influence the direction to change the underlying foundation of how mobile network is built today and sustained towards a path of software upgrade ability towards the future building a foundation that does not depend on hardware a foundation that really is built on a merit of elasticity resiliency and cloud in every component in this network something that I always felt and believed that we resisted this change for different reasons whether it is readiness of technology readiness of organization and so on so I glimpse on the journey in Rakuten mobile as we entered you know just to give you a perspective we haven't been in this business for a long time in the mobile side I joined Rakuten in June 2018 and when I remember when I came to Japan I within the mobile division there was about six people part of my excitement when we talk about and when you hear in the industry that Rakuten was able to build a fundamentally new platform architecture because it's a green field I say you're absolutely right but what is the definition of green field it's about green field as in mindsets mindsets of people and organization that I needed to have in order to build this dream and and yes it is absolutely a crazy idea and the idea was to build a network that's completely disaggregated taking the most complex workload and starting with an area that I really would tell you I am so glad that at least Rakuten now is opening up the debate on this industry around open ran and the importance of open ran to our evolution so we went from you know starting from April 2018 when we were aboard the spectrum we launched in less and less than two years and build our backbone new platform architecture and our cloud in a record time now how we've enabled this is having an absolute fundamental belief that we needed to achieve the following four big pillars in our platform build out number one and this is the most important thing for this industry the radio must get disaggregated this is no longer I in my opinion about a dream this is more of a reality it's just was too important to push to get disaggregation on radio of course it has an impact on cost and it was significant impact in both alpics and capex but I think a lot of people will realize what I already see and what my team see is the biggest impact of disaggregation clarification of the network is what happens in the third pillar I am I call this maybe in the telecom industry we need to bring sexy back to telecom get us into a discussion point a darned autonomy a network that runs on its own a network that discovers issues that is able to heal itself and he unable to grow with that interventions from back office operation and so on so this was really really critical on the strategy level and on open ran you know we pushed into in the early days it wasn't simple it wasn't very easy to convince and even I think I mentioned this before I spent an enormous time trying to convince Nokia in the early days that this is a good thing for them a good thing that to open their interfaces on their remote radio head allow us to do the integration with our open ran software vendor and in fact they did this and I and I tell this all the time this is the carriage that is needed in this industry to move forward but we quickly diversified the supply chain and the supply chain was about hardware massive diversification of hardware sets from macro mini macro small cell enterprise femto and a highly innovative completely all-round compliant 5g platforms for both millimeter wave and sub six and and what a world of a difference now we have seen our open ran platform architecture was not built just to achieve only capex reduction but the critical thing that happened with it is a philosophy and an idea that our licensed radio products need to behave exactly like wi-fi access points they need to have zero touch provisioning they need to have flawless automation on how you configure them we need to abandon legacy organizational structure and move forward into a new world in which autonomy takes the essence of the DNA of how you run and manage these networks and in the context of doing so we realize actually the minute that you have achieved mass scale automation now this is where the elegant uses cases starts to come about when you disaggregate and from day zero even on lte especially on radio you have the separation of cu du and then 4g and packet core you implement user plane function disaggregation we decided to build the world most advanced and largest mobile edge data network in japan and to my knowledge i think this is actually the largest across the globe we've commissioned in less than 14 months over 4300 data centers that are not manned fully autonomous on how we configured our racks how we installed the workloads how we moved from hardware based bbu to a software based baseband running on our far edge data center all of this is connected through a massive optical backbone that supports photonic mesh this is where the future use cases for 5g as we look at edge this is what i am looking forward to is what are the undiscovered use cases that are going to take advantage of this massive scalable edge architecture and the ultra low latency that we will deliver to the application layer and in terms of we also got asked well how the network is running in japan is it really a network that only works in rural areas is it a network that scales can it support the capacity that's needed i mean obviously i would tell you tokyo is not a rural area tokyo is a very very dense area we carry a significant capacity in our network and we started with a premise that everything in the mobile network must be built on an ipv6 architecture since inception we have almost zero outages on services due to the reliability that we implemented on the core we have now had the largest network functions deployed on our cloud platforms with over 800 core network functions fully virtualized fully redundant uncompletely orchestrated this is unique the metrics and performance kpis that we see i think these are world-class if you look at drop call rates and even better the rcs which in rocket and mobile network today north of 87 percent of our traffic passes through our rcs application because we have accepted that voice is an app so we abandoned this concept and philosophy that we need to fight for traditional legacy services and innovative and enhanced messaging and and voice on the cloud the entire mobile network has been built on two hardware types completely commoditized and designed for ultra simplicity and driving also automation and better and easier inventory management uh in our platforms on radio i think this is an area where i as i mentioned before demystifying the secrets of radio is really vital and and this wasn't easy i i'm not going to kid anybody that to go to open ran in the early days for racquetan was a nightmare a nightmare to get the software to work at scale it is very different to say that you could do one p o c in the lab and connect a few devices and say that you have accomplished a great job but what we wanted to prove that you could connect up to 700 devices simultaneously and the system can scale and the software can behave and the underlying cloud and the changes that we have to do to make the cloud telco ready believe me this is not trivial there's a lot of things that we have done to make sure that the software uh workload of radio is running at scale and delivering the performance and reliability and then the parts that i am personally so thrilled about i mean of course open ran and the core and the cloud are all great and this is the parts that i get thrilled about is now what is the undiscovered opportunity that uh that we could unlock this is about all automation you know moving into relentless way of how you need to run and manage the network it is truly very very easy for me to tell you that on the core network today it is i don't have hardly any operational people i on the radio side we started now saying maybe we don't need field technicians because i removed all of the electronics from the site and i managed them in an unmanned data centers with enough resiliency and healing on the underlying platforms so the automation is so critical and so important but it's an end-to-end story and it's not about a function into the network we spend a considerable amount of time building the systems that is is carrying today racket and mobile network and the objective for all of this is just about customer delight problem discovery problem resolution in a much more agile manner than it ever has happened before in this industry um one of the key things that we have done is the investments that we have done on our own what we call racket and communication platform is a creation of something i haven't really shared this with with anybody so i'll share it on on this forum this is what we will now announce to the world in uh in june of this year is the availability of the world first enterprise telecom app store the entire management plane that anything you need from inventory management performance management configuration management vnf cnf lifecycle management uh digital workflow incident management every tool set could you imagine now it all resides in a symbol microservices non-manolithic architecture completely cloud native to minimize complexity and allows you to drive more productivity and efficiency into this network this is something that we are really really thrilled to share and start to really discuss with with with perspective customers of racket and outside of japan to materialize the stream of an open platform towards os s b s s and orchestration and uh running and managing today into morris networks and then one might ask okay well so what great technology what does this really mean to society what does it mean to i think all of us have talked about it in one industry forum or the other about the number of unconnected people across this world but the importance of connectivity to society transformation so i'll share with you that the underlying cost structure of racket and mobile network i think it's offering us an unbelievable opportunity to completely reimagine how we offer our customers our rate plans data consumption plans etc we did something i believe has never happened in the world and we offered now the complete connectivity up to one gb for free for life this hasn't happened at all but in order to achieve those metrics you know we really thought very carefully about how we're going to manage our cost structure how we're going to manage our revenue and make sure that we have a sustainable business as well so if you see this and if you walk into a racket and store today both offline and online there is only a single plan there is not multitude of plans one plan that starts with free if you consume less than one gb you will not have to pay anything and by the way this includes international minutes as well and international calling and and and so on and then you have a ladder that that that that you could pay depending on data consumption model but at best at war sorry if you go to an unlimited model it's about 27 u.s dollars for a completely unlimited plan in rakatan network and for us i can't tell you that the approach that we have taken here i wish i had a bit more time to tell you how we negotiate with our vendors now this wasn't easy every vendor that came and every partner that came to rakatan i spend enormous time of energy to educate them about the importance of open source and i have classic examples that we have taken proprietary software stack and completely made it open source into our platform architecture so the first rule of engagement with rakatan can you make your platform open source i will embrace you help you support you and show you path to revenue and you know lastly i'll end on to this i mean obviously our vision is to revolutionize this telco industry we want to help we want to partner and we want to be open completely open and collaborative on a new world and a new approach on how we go not just only in Japan but provide lessons learned and provide valuable insights for anybody that is willing to think about a world that is driven by software not driven by hardware i think this is the value that we could bring into the table and this is you know our hierarchy and dream is the realization of what we call our rakatan communication platform targeting four pillars one is about cloud and connectivity two is about network functions three is about intelligent operations and fourth is about digital marketplace experiences so i hope this has been informative for everybody and maybe i'll hand it over to you just to see if you have any questions or answers yeah no this is this is amazing tarik i mean i'm sure you're going to be bombarded with questions but let's take a few right away how big are your thousands of edge data centers typically so so we put about one to two rack and it could be the maximum is up to three racks i basically host anywhere between the first true far edge workload today is my viran workload and and so we host right between 30 to about 98 sites or so on on our edge platforms there is 4300 and a little bit higher than this today that we've deployed in japan and by the time we're done maybe we'll we'll end up at around 4800 i know this number might seem quite a bit but there keep in mind these are completely unmanned they're autonomously managed and you know the use cases that we're going to look for is what can we enable next like maybe about our own cdn to be deployed in a few of these locations because the distance between the site to the far edge is only 30 kilometers is not that far yeah okay okay got you any breakdown on ipv4 v6 traffic yeah so so this is a this is a really interesting thing i mean when i when i came in here um this was more of a cultural issue you know to educate everybody that part of the contractual agreement that i even did with my own vendor so the traffic inside my network 100 ipv6 so i mean just imagine this fundamental transformation and thinking i mean i i'll tell you one one quick thing even when we started our own rcp journey as you know ipv6 on kubernetes world was not even a g8 yep so to go through these basic and change that today we drive right about 70 percent of our traffic is going through ipv6 internally 100 percent and externally we're keep driving this and and i call it uh you know i have an i have an external promotion department that encourages the utilization of ipv6 including with third-party companies and third-party websites to encourage them about migration to from ipv4 to ipv6 very good i know we have several other questions will probably take them uh take them offline uh i think i'll take the last one here which is as you move to obviously you've you've been a big supporter of open source for sure um and and and then in terms of disaggregation obviously open ran is is or or an and you know things are important i think we are seeing the world telecommunications providers also move to an open source world which is very encouraging uh and they're dragging standards along right at c 3 gpp etc so what's your view on sort of open source and open standards right as as they move forward you know this is uh so so first of all i i will tell you standards are extremely critical um in terms of their relative importance to bring the industry towards you know formality on the creation of the solution but i but i also will tell you i i have a maybe a slightly different point of view it will always be factual thing that standards will take time to evolve and develop and now what i would suggest for people to consider and think about is it doesn't mean we as enterprises should stop evolving it doesn't mean that we shouldn't stop exploring partnerships and models of collaboration i think for me right now it's becoming clear the beauty of that open source is just unreal unreal when you have the whole world working with you to collaborate to co innovate to co develop and in certain contexts it's okay if you are slightly ahead of standards and you go back in and try to fix an address but i i much rather be agile than just to say i'm just gonna wait for years until the perfect thing comes across so i am i am so thrilled about this community aspect i feel like there's a family and the family understand the value of what they're all contributing towards and you know surprisingly i'll tell you about this i i we started showing the vendors how to make money out of open source and they're liking it it's a great model it's a really win-win for everybody absolutely i think that is a great statement to end and we really thank you to rick thank you very much thank you for taking the time thank you thank you really appreciate it all right okay uh for the next presentation we have uh you know one of the largest uh service providers in china uh china telecom and we have uh sun qiyong uh who is in the sdn research center at china telecom and uh they are thinking ahead on how to move not just automated network but automation into intent based so it is a very fascinating journey that you know i know i am very excited about but without any further delay please take it on okay thank you uh thanks api yes and it's my pleasure to take the presentation here um okay so let's take uh start our uh slides next please yes in in this presentation i will just focus on our cloud network convergence this is our strategy for china telecom and i will discuss about the scenario of intent based network and also our current research and the development of ibn network okay next please yes this is uh i think this is uh very familiar for all of you and we really have increased this amount for cloud network convergence uh from our customers and our industry uh from the point of view of cloud network convergence i think we have different faces for example uh in the first phase uh our customers supporting system and some applications will transferred to the cloud or or application will be cloud native so the customers will need different connections from their side to the cloud they need different kinds of connections like high availability and high quality and low latency but to the next phase and when the application and the requirements become more complicated and they will need a lot of different uh requirements so we need to introduce a lot of different intelligence like ibn i think and after the third phase uh then the cloud will be just a part of the network and the network will be just part of the cloud uh there will be just in in one uh integration so in that part the network is just a lot of network devices will be built uh cloud native so in that phase the cloud and network will become more complicated and we need to introduce a lot of more intelligence in our supporting system so next phase yes and this is the architecture of our overview of cloud network convergence in our city 9 to 2030 this is a white paper published by China Techcom last year and in this whole architecture there are many three parts the first one is we introduce a unified cloud network infrastructure so in this structure we will have cloud and some and we will have different kinds of network and for the next next one is we will have heterogeneous resources for example the 5G, 6G, internet obscene and the fixed line and satellite we have different heterogeneous resources and they will also to be connected to the network and to the cloud and the most important i think what we really need and this is the missing part is the cloud network operating system this is a cloud cloud network a very intelligent brain and that will be very important to support such a large and the huge network and the cloud and to to provide our customer very good our service okay next please so I will discuss some details about the cloud network operating system actually in this in this architecture in the in the main part this is the network cloud operating system the operating system will need to abstract and manage and orchestrate various resources including the network of the the cloud in a unified way and also need to support cloud native development environment and also we will abstract different resources from the network and from the cloud and open it to third party applications so different people can use these different capabilities and there are some major features here like we we need to have the digital wing to simulate and monitor the real-time status of a cloud cloud network operation system and we also need to have self-feeling capability to have the self-adaptive resource scheduling capability and the of the very important thing is to have the intent based engine the internet based engine is to collaborate all the different parts of the network and using the analysis to control and simulation and to make the whole network a whole system very intelligent and automatically okay next please this we will so next we will discuss more about the intent basically intent based network is a self-during network we need self-during if we do not need to introduce some more mandatory works so it is an intelligent worker can automatically convert and verify and deploy and configure and optimize itself so it is very intelligent and it will run very it is very reliable so it will automatically solve abnormal events to ensure the network reliable and it will achieve the target network state according to different application requirements this is this ITO architecture of intent based network okay next please so in our network we will see some scenarios and the real scenarios for intent based network this is a cloud and network convergence scenario and we see different applications will have different SRA requirements for example the game and some security applications will need low latency and low latency data plane so they will need to choose a part that is the low latency but some industrial will need high reliability and low latency plane so they will have different requirements and also some monetary and backup applications will need high bandwidth so we need to figure out the different applications needs and give their appropriate path accordingly next please and also this is a very real scenario of what we are deploying currently this is cloud edge and end collaboration in 5G and MEC actually we can see we will have a lot of connections and the applications and the content will be in different parts of the network so actually the network itself will become more complicated if there is a problem so you need to figure out which part and you really need a high intelligent system to help you to monitor the whole application the whole service okay next please this is currently our research and development of intent based network actually we have done some research work in the next in the past few years but currently we think really we need to bring the intent based network to the industry so we are doing some project in on-line and we have already have a requirement in on-line release age and we will do the first step in 5G slicing this use case and introduce the intent based network in the on-line architecture okay next please oh this is more detail so if you are interested in it I think maybe we can take take it offline and we can discuss more in the organizations in standardizations okay next please and we also also set up develop foreign in the in February and as this is a special session and we have different speakers from different companies and we really thank you very much to and we discuss a lot of this how to how to implement and in different in on-line and how to modeling and what's the framework okay next please and this is our overview of intent based IBM model based on on-line project in on-line actually we think this is a very good choice to implement IBM because it is still it is already very automated and it has a controlled loop so basically it can match the IBM's requirement very very perfectly and so we firstly we will introduce some nature language processing in yo-yo I and this will translate the intent automatically and after gets the intent of we will set up the monitoring monitoring model in DCAE and using the closed loop we will do the monitoring we will do the policy and we also need to set up as a country configuration into our controller but here we have a missing part like the digital wing here we need to verify the intent in the digital wing to verify the the change will be good or not okay next please so this is what we have already done in on-line in release age we have done the yo-yo part which is to put the intent into the from the users like text or audio then we use the nature language processing to map it into different complicated parameters for the network so the users will be very simple to use the application but the language can network can still understand the language okay next please and next we will also add some closed loop in with IBM in on-line like when after after gets the intent from the yo-yo I we will set up the monitoring we will after the modeling in DCAE and also we will optimize in all OS to change like some resources and to do policy and SO and using controller to make the things work together next please okay as you see we think on-line maybe we can do 80 percent of the IBM work in in the IBM architecture but also we still need some other help and some other collaboration with with some of the project in our iPhone for example we maybe we need some analogies and application analogies and the data analogies to do more intelligent from the accurate standard etc and then maybe we can also collaborate with ODL and the open switch to to process what what intent can can do in the whole closed loop okay this is all what I share with all of you and thank you very good excellent excellent thank you so China telecom's leadership in intent based is absolutely well very welcome by the all the global players in open source right as you know on-app open daylight acumos has a global developer forum and everybody participates and you know this is your leadership is very well very well taken thank you very much thank you thank you and and with that I think we are going to take a break we would love to serve you coffee and give you cookies but you know you're on your own given the virtual event we will be back exactly at seven o'clock pacific I would say 11 a.m china time so please you know get some coffee and come back right on time I know we are on a roll we have Baidu we have ZTE we have you know China mobile going to talk right after the break thank you