After successful completion of the course, students are able to plan, optimize, design, build and operate a new generation of networks that, at the one hand, differentiates in using rather different implementation technologies that are used for access and traditional networks compared to those used within SDN-Twin networking, and on the other hand, this network generation distinguishes between end-to-end flows being the data traffic clients and the networking part itself. All these flows can generally be considered as a digital twin flow since all network protocols can be regarded as a transmitter-receiver twin connection. Since signal generation is considered as the real-world original, transmission should be defined as real twin and receiving as virtual twin. A duplex connection will be two opposite-direction twin connections, broadcast is one transmit-twin and multiple receiver-twins. In case of anycast at both sides an adequate number of the two twin types are present.
Program your own network, take a twin and win. This sentence will soon wake up network providers and will very fast grow in importance as the digitalization process will drive up. In the near future, many different kinds of physical and logical networks will be built for numerous purposes. In this competitive game, network providers will be the biggest looser, while service providers might profit slightly.
SDN-Twin implementations resemble contest game playing either as customer flow management of the network with adaptive learning or as scheduling and forwarding in switches and all cloud units. SDN-Twin physical networks are live-stream simulations providing full operational transparency. Each real network component build its own visualisation segment that is delivered in an organized way due to m-cycles to the coordinating virtual twin in order be plugged into the right screen positions or can be individually observed in full process dynamics. RDMA will guarantee the best delay and green operation performance. It forms the internal addressing scheme and data is delivered in parallel paths. TCP-Twin as one-hop memory-to-memory protocol guarantees byte sequencing and correctness. It operates with receiver credits or pacing and full capacity flows. Parallel path might be of different length, but the transmitted load volume is such that all parallel transmissions are completed at the same time.
SDN-Twin is a graph-structured extension of digital twins to become networking model and its topological and structural mirrored physical reality. Realized, this twin system is in fact a local or geographical distributed computing system with an embedded network and fully integrated edge and fog processing units, all in homogeneous hardware-based software technology. SDN-Twin can be used in two different ways. Either it can be used to visualize and analyze the model with all its dynamics by just injecting prepared recorded or artificial real- time flow arrival patterns into all the model sources, than the model structure will transform this real-time pattern according the model characteristics and that flow transformation can be observed by network performance feedback from the physical network. This is revolutionary for all levels of education as well for research and development. Or it can be used to visualize and analyze the performance of real-live traffic flows within a physical network structure where the model controls the traffic behavior, including congestion and network failures, and again all physical network nodes report the network status to the model, thus providing intrinsically complete operational transparency.
SDN-Twin unifies by programming the usage of computing, software and transmission hardware technologies to obtain efficient homogeneous and operational transparent digital-twin controlled systems of local or geographic distributed networked units. Digital-twin technology inherently incorporates that each networked system is purpose-based planned, optimized, designed, analyzed and software tested before with the generated installation lists the final physical network is build-up. After physical connection testing, the zero-touch network installation should immediately be operational. SDN-Twin packetized networks are implemented as distributed computing systems by extending internal transmissions in chip units and on printed circuit boards to include any existing external transmission equipment, thereby achieving and homogenous technology that also can include all cloud units (data center, edge, fog and haze), access units and peering points) as complete computing system operating over a geographic bounded switching area. The network segments can be stacked, concatenated forming three-dimensional networking structures that also can be and also plugged into existing networks. All Internet addressing schemes are maintained. Segments consist of three networking parts: Universal networking infrastructure, L3 virtualized and customer flows (traditional, cyber-physical, point-to-point, broadcast, multicast, anycast).
SDN-Twin networks operate with RDMA (remote direct memory access) over any distances, memory pooling and structuring, industrial computing standards: PCIe, NVME, CXL, RoCE, InfiBand, processing units: CPU/GPU/DTU,TPU/FPUs, memory units: DRAM, DDR2....DDR6, persistent, non-volatile, chip systems: NoC/SoCs/MPSoC/FPGAs, Transmission: Ethernet, WLAN, WiMax, LTE, G5/G6, LoRaWAN, copper, fiber, WDM, radio, Satellites. Underwater.
Principles
(1) SDN-Twin (software-defined-networking with digital twins) is an universal networking methodology that covers telecommunications, industries, cyber-physical ecosystems and networks characterized by logistic management, strategic operations, social contacts, research, development and education.
(2) SDN-Twin was designed exactly for that big business and can generate tailored network zero-touch designs up to installation and testing in a pipelined manner.
(3) SDN-Twin is an open well-known and established software technology running on computing power interconnected by transmission links.
(4) SDN-Twin runs on a distributed computing platform forming the network infrastructure that also may include cloud systems with edge, fog, and haze computing.
(5) SDN-Twin allows any body to build networks and to become a Hyperscaler in the same sense as Google, Amazon Web Services (AWS), Microsoft Azure and AT&T NetBond.
(6) SDN-Twin operates as network segment on a bounded switching area with a geographical size depending of the processing intensively. Segments can be stacked and concatenated.
(7) SDN-Twin operates under well-known bidirectional digital twin control adapted to networking with one computing unit (brain) acting as virtual-twin coordinator and a number of computing units (body) acting as real-twin switches interconnected by transmission equipment. In fact, this is a hyperscaling network architecture.
(8) SDN-Twin as universal network carries two types traffic flows (internal control, customer data flows) on QKD (quantum key distribution) secured routes.
(9) SDN-Twin inherently exhibits transparent operation
1) New insights about SDN-Twin (software-defined-networking with digital twins) will soon make headlines around the world. A new dimension in quality is reached by applying digital twins in universal networking. In addition, SDN-Twin is a distributed computer system with a physical network and cloud systems integrated inside. There can hardly be found a better solution to build networks. By physically separating network infrastructure and customer flows, universal physical design with only processors, memories and transmission systems combines the best achievements of two strong industrial forces, thereby opening an era of automated generation of lean special purpose networks from design to installation logistics. Customer flows through that computer system may be traditional such as in telecommunications and industrial applications, but particularly include all-kinds of cyber-physical systems that rapidly increase in number. In this respect separating the internal QKD (quantum key distribution) secured network infrastructure from the transparently transported customer flows makes this form of universal networking. SDN-Twin enables also optimized physical networking in the dimensions of space, time, power requirements and complexity. In a distributed, physical network, all four factors are the worst because the values are high, in the internal network of a computer they are already significantly lower and within chips they are lowest. In all three realizations, we are always dealing with protocol mechanisms and transmissions. The geographic size of space determines the overall efficiency value and that must be as low as possible. SDN twin networks, geographically always bounded to a suitable switching area, are topologically present twice, namely one virtual and compact in a computer as network coordinator and one real as a geographically distributed physical network with all switching nodes connected via transmission systems. The network coordinator has a complete overview of his physical network in a compact way and it therefore makes sense to first optimally determine the best real path through the network before internal control data embark on its real journey, because all four factors contribute in the physical network in the worst way to the efficiency rating. This is also the right place to recognize and prevent impending traffic congestions at an early stage, to switch on additional transmission lines in the event of capacity bottlenecks or failures, and much more described below. In short, without bidirectional control between the virtual coordination twin and all real switching twins in the network, one can apply traditional SDN or the new SD-WAN technology successfully, but one can never reach the outstanding properties of SDN-Twin. At this point it should be pointed out that geographically larger networks are formed as a cellular structure with a seamless transition. The size of the cells depends on the traffic density. Furthermore, it is also important to recognize here that there are two types of traffic in the physical network, namely the network-internal control traffic, which could also be called the network twin traffic, and the customer traffic itself. SDN-Twin is not just a network system with the working title "Program your own network, take a twin and win" but can be used universally, either traditionally or for the new trend of diverse cyber-physical ecosystems. All routes in an SDN-Twin physical network are QKD (quantum key distribution) secured. This is a symmetric encryption over any transmission system and has nothing to do with quantum transmission itself, but such protocols are designed in such a way that the encryption can never be cracked, even not by a quantum computer.
All applications of digital twin designs have the common property that all details and aspects are thoroughly optimized and analyzed before being realized. This is also the basis of SDN-Twin as distributed computer system on a geographical bounded switching area with a single virtual twin as coordinator and a number of real switching twins to forward packets. In this environment two traffic theoretic methods are used. For the coordinator the operative design is based on flow performance provided by traditional circuit switching theory started by Erlang in 1915 for deterministic telephone circuits that is extended by the properties of packet flows resulting in smooth, Poisson, peaked and self-similar traffic. This is just a traffic flow capacity difference that has to be considered and as mix defines the required throughput requirement. In the switching twins packet scheduling is required and queueing theory is the basis to achieve the best performance. Both twin topics are part of the design process. This process is the basic unit for stabile operation in a endless life cycle in which each extension will result in an improved operation. SDN-Twin networks are intended to be designed and improved in an continuous manner. This kind of networks must be the dream of each operator then one grows up with a network that is not based on the current roll-out roll-in praxis of today, where equipment learning is an expensive and time consuming factor. With SDN-Twin, extensions can be implemented as gradual steps in form of well-known living networks. In the design process also network topology, resilence and overload scenarios are considered far beyond the expected traffic load and flow constellations. As result, all possible routes and event scenarios are known. The next step is the automated generation of the virtual twin and all real switching twins in software and hardware constellation that finally delivers the foundation of the physical network. One should notify that the design process is achieved and can be activated when ever extensions or analyses are necessary.
Both the network coordinator and all network nodes are in one physical network. Coordinator and nodes are all computer systems in a metallic housing structure with all externally required transmission systems with various forms of implementation. Many commercially products exist. This network mediates two types of network streams, logically separated and strictly cryptographically secured from each other. Internal network control has top priority and can interrupt all other priorities for their external transmission without data loss. The second category is the customer data streams and of course the data streams for which the network is actually operated at all.
This turns SDN-Twin into a logistics system that continuously optimizes the balance between distances, environmental compatibility, safety, economy, efficiency, transparency, performance and resilience. This is because the virtual coordination twin always has an up-to-date overview of the status data of all network resources in its geographically assigned network area using predefined m cycles (m: monitoring). This data is already prepared in the required data structures as an image segment with information data from the individual real switching twins, in order to then be delivered cyclically in a virtual ring structure as RDMA (remote direct memory access) in the data structure of the network coordinator. This is pure modern software technology and can hardly be surpassed in terms of time efficiency. This means that the network model and the physical network are always asynchronously synchronized. The coordination twin intervenes only when necessary.
SDN-Twin is a distributed computing system with an integrated physical network and cloud hierarchies consisting of data center, edge computing, fog, and haze. The dominant technology is modern computer technology with its software technology, which is connected to all existing transmission systems. The SDN-Twin methodology becomes common knowledge and thus network programming, like today's application programming, becomes a completely normal thing.
The networks of the future must be able to cover all sectors. Therefore, SDN-Twin networks are designed in such a way that only the network topology and the traffic-related criteria of throughput, delay, loss, availability, quality of service and security are required. Therefore there is a clear separation between network structure and customer traffic. This is also necessary, because different types of networking and modes of operation must be covered in such a way that networks in the assembly line system can be automated from planning to installation. In the digitized future, you have to be fast, cost-effective and tailor-made. Above all, these networks must also be environmentally friendly and safe. Network, power supply and security is an inseparable unit. Many such SDN-Twin network factories will emerge. The market is unimaginably large and the demand will increase rapidly. The network types are telecommunications, industrial and commercial, organizational, strategic, social and logical. Especially the logical networks as a computer network system play an essential role in all stages of education, training as well as research and development.
What makes SDN-Twin networking so attractive?
(1) Open source networking that will be driven by the newest results of the scientific community,
(2) Simplicity due to separation of network infrastructure and customer data flows,
(3) Universal network design method with purpose-tailored optimization,
(4) Perfect dynamic optimized logistic system that is operated adaptively by digital twins,
(5) Distributed computer system with an inside integrated network including diverse cloud elements,
(6) Networking paths become controlled transmissions within chips/devices and between devices,
(7) The best of two industrial worlds, computing/memory, transmission and networking principles
(8) Network programming as application with well-known languages,
(9) Automated pipe-lined generation from design up to installation lists,
(10) Complete analysis of all available routes and properties before realization,
(11) Using traditional teletraffic flow optimization for the coordination twin,
(12) Using queueing theory models for adaptive scheduling of forwarding in switching twins,
(13) Using twin control to continuously, but asynchronously, adapt the virtual model to physical reality,
(14) Zero-touch installation,
(15) Sustainable because stable lifetime with increased quality and inprovement,
(16) QKD (quantum key distribution) internal network security of infrastructure and customer flows,
(17) Complete operational transparency due to the intrinsic live simulating digital twin system,
(18) On-demand simulative Analysis of detected anomaly events in the original network design versions.
(19) Fast and automated (re)configuration of traffic and security requirements of customer flows.
(20) Early congestion detection with fast load capacity changes and delay timing control,
(21) Fast reaction to link and node failures,
(22) Optimal green network operation by closing and opening routing paths (standby or switched off),
(23) Combination of pacing and credit flow control,
(24) MTM (memory transfer mode) switching in virtual coordination twin and all real twin switching units,
(25) Continuously stable TCP flows with delivery guarantee and feedback notification and resequencing,
(26) Full link capacity flows without losses by using this TCP protocol in a special way,
(27) Lossless operation due to temporary FG-BG (foreground-background) storage,
(28) Controlled network feedback as structured RDMA flows of m-cycles (monitoring) onto the screen segments,
(29) Structured RDMA flows allows physical network status data immediately to be plugged in,
(30) Minimal required protection capacity requirements by p-cycles (predefined),
(31) Simplified network management by detailed visualization and optical guidance what to repair,
(32) Routing through the physical network by dynamic label switching,
(33) Access load distribution for sudden unpredictable traffic load distribution changes,
(34) Computing system on a bounded switching area to be plugged easily into a existing network,
(35) When internal IP addressing should be maintained, freely chosen private IP addressing,
(36) Virtualization of addressing, signalling, mobility and multicast trees executed in real network,
(37) Visual location identification support for faulty equipment repair,
(38) Best operational overall performance expressed by
minimum {Capex (capital expenditures), Opex (operating expenses), Lean} and
maximum {Availability, Performance, Quality features, Security, Green}.
2) The miracle of SDN-Twin software-defined-networking with digital twins
Communication and electricity networks as well as security issues form jointly a mission-critical system upon which modern living and business are founded and made possible. The highest possible measure of understanding in these three fields of human-created constructs are key factors to keep that level of prosperity and to try miseries in our world to relieve by education and showing how to become independent by open network systems. Several algorithms and concepts in scientific literature and well-established implementation methods will be applied to reach that goal. Other important business factors in networking such as service quality, customer satisfaction, operational transparency and the diverse cost factors are taken into account, too. Digital twins are actually a bidirectional control mechanism between one digital computing system that is extended by another geographically located digital cyber-system as modelled implementation of physical reality. Cyber stands for the well-known Internet that interconnects everything. Thus, digital twins interconnect computing with cyber-physical worlds. ISDN-Twin uses this control mechanism to realize a nature-based property that exists since there is life on earth. All living beings can be considered as a twin system that is a real-time communication system to obtain two visions of reality that are coupled. When restricting to humans, one view is virtual coordination and realized in the physical brain, the other view is the real network and realized in the physical body with sensory organs (eyes, ears, taste, smell, skin, pain) interconnected by the vegetative nervous system. Today, this mechanism can be used as bi-directional digital control system. Transmission and computing technologies are the fundament to build networks. Network topologies consist of nodes and links. Networks can be considered either as a collective group of distributed nodes or a collective group of distributed links. Depending of the choice, one obtains SDN-Twin networks in a consistent technology or one migrates and integrates traditional networks to SDN support and digital twins. There many possibilities to implement. Networks are extremely complex and difficult to predict in behavior of traffic dynamics, congestions, operational faults, equipment failures, protocols, security attacks and service requirements. SDN-Twin was born to simplify networking and to become universal, transparent, intelligent and secure. Its implementations is not of all trivial. However, by separating internal network traffic management from external customer flows and using digital twins that always bootstrap the network to the best operating point and cry when the network is underequipped to achieve additional equipped installation at the right places, is a major step to achieve universal commodity networking that works.
This is the right place to explain this extremely effective SDN-Twin balancing mechanism.
(1) All real twins within the physical network provide real status information to the coordinating virtual twin network model with the same topology that adapts its network perception.
(2) This means that all network states in the model and in reality always deviate from each other with a mathematically calculable difference as a manageable tolerance limit.
(3) Based on this information the coordination virtual twin is able to decide to trigger physical network changes or not.
In this context, following sayings reflect a realistic form of implementation:
(1) "Science meets business",
(2) "Program your own network; take a twin and win",
(3) "Once balancing digital twins, then such digital twins forever",
(4) "Digital twins make the difference".
Finally, implementation of SDN-Twin networks, site-internal rooftop-interconnected clouds, their combination and communication with quantum cryptography, being interconnected by radio, electric or optic, will become an universal, pipelined software factory process. To that end, it is very hard work to achieve all these close-related networks types that will be covered.
SDN-Twin topics to be realized
- SDN-Twin networks (telecommunications, industries, business, organizational, strategic, social, logical).
- SDN-Twin clouds (datacenters as well as edge, fog and haze computing units for Internet-of-things).
- SDN-Twin access systems (underwater, terrestrial, sky and space with access via acoustic, radio, wires, optic fiber).
- SDN-Twin devices as hardware realization of digital-to-digital and digital-to-cyber-physical twin systems.
- Communication systems between digital-to-digital and digital-to-cyber-physical twin systems.
- Real-time communication over packet-oriented networks based on digital-to-digital twin systems.
- Real-time simulation of mathematical models such as queueuing systems and networks as well as algorithms and methods with all dynamics.
- High-performance visualization to observe the interactive dynamics of SDN-Twin models and cyber digital-physical worlds.
Important highlights leading to SDN-Twin networking
1970s: SPC (stored program control) which is a computer system for telephone communication centers to set up telephone calls.
- Computing system (with geographic network inside).
1980s: SONET/SDH worldwide backbone transmission network that still is the most important backbone system spanning around the world and the biggest geographic distributed byte-machine that exists and operated by many operators. Network channels for phone and packet flows are setup and teardown.
- Distributed computing system (with geographic network inside).
1990s: Optic/photonic multi-wavelength networking (transmission, switching, processing, monitoring).
- Network (with geographic-distributed computing systems inside).
2000s: Transformation to packet-oriented networking.
- Network (with geographic-distributed computing systems inside).
2010s: Start of SDN (software defined networking). Start of digital twins in nearly all business areas.
2020s: SDN-Twin is the coronation of all these activities that span mathematics, informatics, telecommunications, computing, business and security to realize open-source products based on international education, research and development in hardware, software, implementation and operation.
- Dual-plane distributed computing system (with geographic network and clouds inside)
SDN-Twin is a special tailored network or a special tailored cloud that includes floor-shop processing power.
In both cases, SDN-Twin is a two-plane processing system based on software-defined networking with digital twins.
SDN-Twin is a two-plane distributed computer system that as network includes a geographic bounded distributed physical switching network and that as cloud includes a virtual switching network with physical rooftop optical interconnect on top of the computing cabinets.
SDN-Twin was born to simplify networking and to become universal, transparent, intelligent and secure.
Simplification because of introducing of a bi-directional control loop between network model and the corresponding physical network. Digital twin communications interconnect computing and their cyber physical worlds. Network model continuously adapts to reality. Simplification also because of separating networking operation such as routing, resiliency and congestion control from application and service network data flows. Universal because the model twin is a virtual network based on traffic-theoretical methods that only controls and adjust transmission capacities and manage delay management and exceptionally events. Transparent because all physical network switch twins report their measurements to the virtual model twin for display operational quality. Intelligent because processing power is available in both planes of the SDN-Twin system. Secure because network control flows and customer network flows are, even when using the same physical network, network designed secure by. And on top each customer flow is upon entering the physical network screened for protocol and security anomalies.
3) History and description of digital twins
Digital-twin systems are real-time, bidirectional communication systems to obtain two descriptions of reality, one as virtual twin to describe a coordinating unit and one description resulting from all real twins within the physical system, and thereby providing a coupling between both descriptions. Applications determine the time target of this coupling. In this manner, couplings between digital processing systems and cyber-physical systems or worlds can be established as control systems. The word cyber implies that a digital processing unit that describes the physical environment exists. Therefore, the control mechanism operates between digital twins with several verbal twin combinations such as (virtual, real), (logical, physical) or (coordinator, switches).
Twin mechanisms exist since there is life on earth, but to become a digital twin mechanism, digital computing units were necessary, so that this twin concept really was started by NASA in the 70s to realize the first moon landing on July 20, 1969 with Appolo 11. The name digital twin is due to Michael Grieves for lifecyle management and John Vickers of NASA in 2010.
Twins: All living beings can be considered as a twin system that is a real-time communication system to obtain two visions of reality that are coupled. When restricting to humans, one view is virtual and realized in the physical brain, the other view is the network and realized in the physical body with sensory organs (eyes, ears, taste, smell, skin, pain) interconnected by the vegetative nervous system.
Digital twins:
(1) Weather news: Weather prediction is based on a increasingly improved detailed model based on providing an huge amount of real world data. This is an example for a digital twin system that cannot influence the weather, but can inform to act accordingly.
(2) Route traffic news: By continuous traffic data measurements and additional information of road construction, accidents and congestions, radio broadcast and traffic navigation devices in cars might influence decisions of drivers to take another route. This is an example for partial control.
(3) Transportation: Trains, trams, busses, trucks, planes, ships must or will likely follow the directives of the control centers.
(4) Construction industry: In this example and all following digital twin examples, the well-defined and stable lifecycle as well as the structured way of operating progress is driving motivation for using digital twin methodologies that also included a digital twin thread to document different installed versions. As well-known example, building a special designed house by an architect has been taken. This is an interactive process between customer and planner with visual tools that reveals all customer required details until functionality, design, material and costs agreed. This generally means that with digital twins, all projects for all business sectors can be planed, optimized, designed and preparations for building the end product, including its operating features can done and evaluated before any constructing realization starts. In this construction example, the planned house will very well match the final realization of the house, whereby of course small changes can be corrected during its building time.
(5) End-to-end protocols: All these protocols can be considered as digital twins, where the transmitter as original flow initiator is the real twin and the receiver is the virtual twin with the task to correct data flow errors.
(6) Other Applications: Industrial, manufacturing, agriculture, healthcare, hospitals, tourism, emergency services, cities, towns, villages and rural and sparsely populated areas, mission-critical infrastructure (utilities, electricity, pipelines, submarine cables) that all can be realized as SDN-Twin networks, including clouds, edge computing, traditional network access and coordination networks for moving networks such as cars, trains, ships and airplanes as well as the special cloud application of model visualization of dynamic effects for enhanced understanding in schools, colleges, universities of applied sciences, universities and research laboratories and research centers.
4) Quantum entanglement twins and SDN-Twin control digital twins
The topic "SDN-Twin: Software-defined networking with digital twins" has gained another facet. There is something in common here with the Quantum Twins from colleague Em.O.Univ.Prof. Anton Zeilinger who has just received the 2022 Nobel Prize in Physics. Quantum entangled twins and control-oriented digital twins enable the same effect, so to speak, namely state binding. The entangled bond with two exact states is hard, with digital twins connecting a cyber-physical world and computer is soft, but through state information from the real world, the corresponding model can continuously adapt its parameters and so the relevant states only soften in a mathematical certain error range from each other. Quantum entangled twins are the basis for highly secure encryption methods. Digital twins and their regulation are comparable to sensory organs and the central brain. On the one hand, SDN-Twin enables the networking of all pairs of twins from cyber-physical worlds to the central coordination system and, at the same time, SDN-Twin guarantees that all customer data streams are continuously verified with the highest quality features.
SDN-Twin network edge partnership opportunities
Hyper scaling by virtual networks to manage an embedded, topology-matching physical network with peering points and cloud hierarchy of datacenters, edge-computing, fog and haze units. The concept of SDN-Twin to host enterprise and user networks as self-similar SDN-Twin networks.
SDN-Twin can turn telco's into powerful and vastly superior geographically distributed Hyperscalers. Well-known Hyperscalers are Amazon Web Services (AWS), Google Cloud Platform (GCP) and Microsoft Azure. These cloud resources are easily accessible and scalable. With the SDN-Twin networking methodology as a network within a geographically distributed computer system, each switching node can also become a local cloud cluster. This means that all XaaS services are integrated directly into the telecommunications network in many "as a service" variants and can be marketed accordingly. Therefore, the economic losses caused by today's cloud Hyperscalers and SDN-WAN competitors can be slowed down. Network internal cloud clusters are also the superior solution due to better interconnecting networking and the associated availability. As known, SDN-Twin will play an important role in many physical networking types (telecommunications, computer, organizational, strategic, social, logical). The digital twins are the crowning glory.
SDN-Twin can be implemented as a virtual cloud network in a single box. From small to mega large. The methodology remains exactly the same. There are various options for implementing the internal transmission. With MTM (memory content transfer mode) switching is realized. Cloud computing is an essential component in all SDN-Twin network types (telecommunications, computer, organizational, strategic, social, logical). A mixture is also always possible. Each cloud has a virtual coordination twin and the physical real twins are the entry ports, because that's where the real traffic comes into the cloud network.
Hardware: DDRxRAM, NVM, CPU, GPU, DPU, CXL lanes, internal optical connections or Ethernet backplane.
Software: CXL, MVMe, CUDA, RoCE.
Memory: DRAM, DDR2... DDR6, persistent, NVM (non-volatile memory).
• PCIe (peripheral component interconnect express),
• CXL (compute express link),
• NVMe (non-volatile memory express),
• CUDA (compute unified device architecture),
• RoCE (RDMA over converged Ethernet),
• CPU/GPU/DPU chips (universal, paralleling, routing included).
Public network or service operators can also use the SDN-Twin networking method with their existing network equipment. Furthermore, one should not get the hyperscalers on board. This is just a loss of business. It is essential that routers and switches do not act independently and in a controlled manner. There must be a virtual coordination twin and a set of arbitration twins in a geographically limited area. Within the framework of SDN, protocols such as Openflow and PCEs (path control elements) sufficient knowledge in the telecommunications sector. Of course, one can also achieve the same goals. However, one needs to know mesh armor, and there may be multiple manufacturers. SDN-Twin is neutral and does not require any knowledge of the L2 to L4 layer protocols and their properties. The twins always ensure that the coordination knows the network status. This will of course also be the case in the above Telco methodology. Only in SDN-Twin you don't need to know anything about it, because only data traffic is routed through the network. With SDN-Twin, almost anyone could build and operate networks. That's a big plus.
SDN-Twin networking is a universal, traffic-theoretical and geographically distributed realization of a computer system through which customer data flows with quality guarantee and the highest security measures. Customer data streams are all physical transmissions that realize the various networks for telecommunications, computers, organizational, strategic, social and logical. Thus, it is essentially a manager for physical network resources with a digital twin control mechanism between one virtual coordination twin and all real switching node twins. The physical network provides real status information and the network model with the same topology adapts its network perception. This means that all network states in the model and in reality always deviate from each other with a mathematically calculable difference as a manageable tolerance limit. As in all digital twin systems in all industries, all network properties and the software for the model and physical network are tested and generated on a computer as a preliminary phase. The software is then loaded onto the hardware of the twins (once virtual, several real). After physical network installation with a connection test, the network is ready for operation without configuration.
The logical networking within an SDN-Twin computer system as a processing accelerator results in an internally configurable, physical network for real-time visualization of any complex models of all kinds, including cyber-physical worlds. Professional visualization of dynamic effects is the fastest way to learn permanently. SDN-Twin visualization in clouds could become an important Internet application for schools, universities, technical colleges and general for all business areas and interested parties.
SDN-Twin is a data traffic infrastructure that is set up in terms of hardware as a geographically distributed computer system with various types of processors, memory hierarchies and transmission technologies. When it comes to software, open rules apply. Customer data streams are not part of SDN-Twin. This is what makes it so universal and task-oriented. The full concentration is directed towards traffic theory tasks and modern intelligence topics.
5) SDN-Twin history
1970s: Computer system SPC (stored program control) in circuit-switched telephone exchanges.
1980s: Byte-interleaved machine SONET/SDH (synchronous digital hierarchy) and all extensions.
1990s: Photonic networks with WDM transmission, optical amplification, switching and processing
2000s: Transition to packetized networking, performance control and network resiliency.
2010s: Ultrafast structured-memory switching by computer control using NVMe, PCIe and RoCE.
2020s: Ultrafast structured-memory switching by SDN-Twin control using NVMe, CXL, CUDA and transmission
SDN-Twin networks take Internet addressing into account and virtualize signalling, mobility and multicast trees.
Entering in edge-cloud and data-center cloud processing is a natural step because both network parts are build in a homogeneous manner.
SDN-Twin networking concentrates on five modular parts:
(1) A traffic-theoretic optimized network model for transparent customer data flows that
(a) after being thoroughly analyzed and
(b) being implemented as software and
(c) loaded on bare-metal switches having commercial processing units and being interconnected by any kind of transmission technology, automatically becomes the real physical network with full transparency, thus, providing continuous verifiability of performance and service quality and intrinsically enabling further network or operational strategic extensions,
(2) Traffic characteristics and performance requirements of internal control and customer data flows,
(3) Scientific methods,
(4) Software and hardware implementation, testing, equipment installation, network operation, and security,
(5) Understanding the diverse customer cyber-physical worlds to define customized twins and their traffic characteristics.
Scientifically, SDN-Twin is a control system between a network model and a topology-identical, real physical network. Upon receiving real network status data, the model can update its network status view and can take actions when needed. Structurally, SDN-Twin is a two-plane distributed computer network on a geographically bounded switching area in which the network topology is viewed in two different ways. Virtual with one computing system that observes the complete switching area and therefore this virtual twin supervisor is a route navigator, operational coordinator and emergency manager. Real with several computing systems operating as real twin switches interconnected by transmission links, thereby forming the physical network.
SDN-Twin is a special core network technology that only uses commercially available processing units, memory hierarchy and transmission technologies. That enables building open networking for everybody that is appropriately skilled in the topic as well as in computing software and hardware. Access technologies with all applications and services as well as edge-computing clouds or any kind of access networks incorporate the core flow clients. Their technology is a mixture of all what exists. In operation, the real twins report what is happening in reality, the physical networks, the controlling virtual twin updates its status view and only reacts when necessary. The most significant property of digital twins in all business sectors is that all what might happen has already been analyzed on a computer system, before a product, process, and in this case, a real physical network has been realized. Improvements based on previously not known issues, can be included at any time.
All this is a historical milestone in lifetime-stable computer-based planning, design and realization of a network traffic-control system for arbitrary business purposes that automatically becomes the real physical network with full transparency, thus, providing continuous verifiability of performance and service quality and intrinsically enabling further network or operational strategic extensions. This is an exiting new way of networking being either organizationally, strategic, logical, computer or telecommunications oriented. All distributed computing systems, one virtual control twin and all real network switching twins, are embedded in so-called bare-metal switches with a chassis construct to attach all required transmission interfaces. In the spirit of SDN-Twin, processing units are programmable in well-known languages and use advanced software controlled memory hierarchy environments.
SDN-Twin networks are machine-learning and self-adapting systems. This is given by the bi-directional control flows between the virtual model twin and all real network twins. This control is not strong synchronized, but just adequate deterministically bounded. The network traffic model manages the resource boundaries and defines limitations and rules. Thus, it defines the operational space. The network twins have the knowledge of the real network traffic situation. Depending of the purpose of the network, the load pattern can be rather stable but load situations may also highly fluctuate. In general, load patterns will change by time and the network model adapts is behavior at equitably and when necessary requires the installation of additional network equipment. Faults or network malfunctioning repairs belong to regular operational activities.
SDN-Twin network design testing is very effective and fast. This is because a universal data traffic theoretical core network as two-plane distributed computer network system is completely separated from all application (digital twin) flows that use the physical part to be carried through. These flows actuate the network status view in the virtual twin coordination plane that checks whether all currently activated resource capacities are sufficient and all the performance objectives are met. This SDN-Twin network structure can be used to run many source code objects of in the institute existing event-by-event simulation programs. This will help to obtain early results before a professional software library has been created. By this step, an accelerated testing of all properties of the final physical network implementation can be done on a universal laboratory test network before the preparation of the real hardware implementation starts. The different network-access client-flows used during test are adequately aggregated data streams between entry and exit points of the bounded switching-area network. These flows are based on different data traffic types (deterministic, smoothed, exponential, peaked, self-similar), characteristics of end applications and services (all generalized as digital twin flows), aggregated as routes through the network) and additionally considering a stochastic variation. Network status information flows from real physical to virtual control and possible bottlenecks can be detected. In future, this will become a general method to perform fast network real-time simulations, thereby revealing dynamic effects and behaving due to live visualization.
SDN-Twin networks are universal, physical networks (telecommunications, computer, organizational, strategic, social, and logical) based on digital twins to interconnect any digital twin implementation of each business sector as client flow. This two-plain distributed computer system is internally coupled by a virtual twin for operational coordination and many real physical twins as switching units that are physically interconnected by transmission links of any flavor. Virtual and all real twin computer systems are embedded in a bare-metal switch housing with commercial processing units and transmission interfaces. This physical network carries two types of data packet flows, (1) internal control having the highest priority and enabling to locally preempt all other data packet priorities and (2) all physical network client flows from applications, services and digital twins in general. Technically precisely specified, these packets are layer-2 data frames. The network topology seen logically within the virtual twin is the same as the topology physically formed by all real twins.
Hardware signal propagation delay is one major difference in the two-plain distributed computer network. Within the virtual twin and within each of the real twins, these delays are very small but as ultra high-performance processing system a extreme crucial issue for designing interconnections between chip-units, interfaces and circuit board very carefully. Considering signal propagation delays for transmission is in networking well-known. This is a physical effect that always remains; propagation delay mus/km for vacuum (3), air (3.3), fiber optic (5), coaxial cable (5,4), twisted-pair copper (6). With respect to the real twin switches, one should notice that such an unit has an embedded a twin system in itself. The virtual part coordinates switch internal tasks. The real part performs scheduling and switched forwarding, including data traffic statistics as well.
With respect application purposes, SDN-Twin universal, physical networks can be split into three groups. First, are these the well-known computer and telecommunication networks. Second, a new class of networks, in which there exists a coordination function (virtual twin) and in order to execute that functionality as network operations, either on an own managed physical network or by temporarily using a variety of public networks. Such realizations are fulfilled by the organizational, strategic and social networks. And third, the logical physical networks as most prominent type will become the most valuable form of SDN-Twin networks in education and science. The main attraction is real-time visualization of dynamic effects of graph-oriented problems. This also means that interesting network simulation models or process able numerous studies, master theses and dissertations, spread over the whole world, can become a living twin in order to possibly reveal up to now hidden effects or intrinsic dynamics. This is a kind of archaeology of forgotten work that gets a revival with additional results and insights due to real-time visualization. Another feature of SDN-Twin networks is easily to become a hardware accelerator in all matrix-oriented problems with or without visualization by using combinations of CPUs (universal computing), GPUs (massive parallel processing) and DPUs (routing for graph-oriented problems).
In fact, one can extend this list to include TPDUs (tensor flow) for neural networks and FPDUs (file flow) to bring all the data needed as a destination-driven, credit-based data stream into the fetch cache of a specific CPU in order to do the Distributed data collection overcome across different memory banks with possible delays caused by memory read collisions. When combined, these become FPDUs. Combined with the physical part of the SDN-Twin network, this becomes a tool for visualizing complex systems as overall dynamics in real time, which helps to understand many effects or effectively perform the SDN-Twin detection of anomalies in logs of customer data streams and security protection.
6) Education in SDN-Twin networking is a professional 2-year course in planning, design, installation and operation of SDN-Twin physical networks and their customer applications, services and their particular business-sector digital twins. This course is embedded as optional voluntary lectures in the education and research environment of the Master Program of the Institute of Telecommunications and as such this special education will be documented as visited lectures, but this networking topic will, part from some credit points, is completely decoupled from obtaining the Master degree. The target of this course to acquire additional expertise and comprehensive knowledge of SDN-Twin networking as fundamental new open form of networking.
This course starts every year by giving the central lecture and two different alternating detailed topic collections of 9 lectures spread over two years, thus forming 19 different units in total covering all aspects that are necessary. Additional practical experience can be gained as soon as the first laboratory SDN-Twin grid-network has been installed and brought in operation. That might already be the case mid 2023. This laboratory is planned to become the international test center of SDN-Twin networks. This special course program can just be visited with of without examinations, there will be a very small number of Master thesis topics written out and PhD students in connection with the various national PhD programs will selectively be taken.
389.209 SDN-Twin: Software-defined-networking with digital twins
389.226 SDN-Twin automated software and hardware design, extensions, tests and installation logistics
389.213 SDN-Twin CPU/GPU/DPU processing with memory hierarchy and transmission technologies
389.217 SDN-Twin clouds, edge-computing and access networks
389.211 SDN-Twin optimization based on mathematics and metaheuristics
389.214 SDN-Twin path search, p-cycles, m-cycles, resource allocation and scheduling
389.215 SDN-Twin performance modeling and data traffic flow theory
389.216 SDN-Twin anomaly detection in customer flow protocols and security protection
389.219 SDN-Twin processing of addressing, signalling, mobility and multicast trees
389.220 SDN-Twin performance management of real-time packetized flows of voice, video and control
389.221 SDN-Twin real-time analysis and visualization of interconnections between cyber-physical worlds
389.222 SDN-Twin deep learning and analytics for physical network control and machine-to-machine flows
389.212 SDN-Twin: Digital-twin applications and services in all business sectors
389.208 SDN-Twin in metropolitan cities, towns, villages and rural plus sparsely populated areas
389.210 SDN-Twin in buildings, industries, farming, transportation and mission-critical infrastructures
389.218 SDN-Twin for twin-based moving networks like cars, trains, ships and airplanes
389.223 Real-time simulation of mathematical models, algorithms and methods with all their dynamics
389.224 Real-time communication over packet-oriented networks
389.225 Design of distributed computing systems with physical network and clouds inside
389.227 Modelling of cyber-physical digital twin systems and their technical realization
Publications
[1] SDN-Twin all-optical routing by software-defined networking with digital twins, Revival of optical burst switching as generalized optical flow switching, World Congress of Smart Materials - 2023 in Barcelona, July 22-24, 2023
[2] Cloud systems with high performance rooftop interconnection in different ways of integrated optics,
AMSE 2023 4th International Congress on advanced Materials Sciences and Engineering, March 17-21, 2023, Vienna
[3] SDN-Twin networking with MTM forwarding and FTM routing in photonic networks,
ONDM 2023: 27th International Conference on Optical Network Design and Modeling, Coimbra, Portugal, May 8-11 2023
