Gruppo TIM
Gruppo TIM

La nostra visione ESG

ESG vision

Crediamo che la trasformazione digitale sia la chiave per lo sviluppo sociale ed economico del nostro Paese e per la tutela del nostro territorio. Approfondisci

Edge Computing is an architectural solution that enables the availability of contents and applications closer to the end customer based on a flexible network in terms of traffic management, access technologies, automation and service creation and represents a central element in the “Beyond Connectivity” model.

Under the assumptions and guidelines described in the paper, Edge Computing represents the opportunity for TIM to gain the role in the value chain by leveraging customer proximity and tight integration of the telco and service components.

Scarica il PDF

Down

Articolo 03 - Edge Cloud ...

2794 KB

Introduction and Context

Edge Computing is an architectural solution that enables the availability of contents and applications closer to the end customer, to meet the requirements of latency, security, resilience and bandwidth; it also allows a more efficient distribution of computational capacity between de vices and network. The development of the Edge framework can involve a Telco Operator, an alliance of Telco Operators and Hyperscale Cloud Provider having partnerships with Telcos.

Current Situation

TIM Network infrastructure
TIM network is developed on a hierarchical architecture with 3 levels of aggregation: Local Aggregator at Local sites (Central Offices), Feeder at Regional sites, Metro PoP at National sites (Figure 1).

cap03-01 TIM Network Architecture

Figure 1: TIM Network Architecture and RTT

click here to enlarge the image

The local and regional levels perform functions of aggregation and transport of traffic. The progressive expansion of IP technology up to Local Aggregators (R-Evolution project) is a pre-requisite for the descent of network and service functions to the Edge. National level hosts the Fixed Core Network functions, the Mobile Core Network functions (in progressive expansion to reach all the National PoPs with the new Cloud Native 5G Core Network) and the service and content distribution platforms, including Content Delivery Networks (CDN) and OTT «Alien Caches». Interconnection to the Public Internet takes place at the International Peering point.
Under the current network structure, distributing the functionalities currently present (or planned) on the national sites to the regional sites, would produce an advantage of about 1ms (see Figure 1) in terms of reduction of Round Trip Time (RTT), which can be considered negligible compared to the latency requirements of most of the available applications. Currently a subset of PoPs is hosting Telco Cloud islands, which are Data Centers featuring a single virtualization technology (NFVI, Network Function Virtualization Infrastructure), specifically designed to meet Network Functions requirements. TIM NFVI is already evolving to further increase its footprint in all PoPs of the national backbone and, depending on market needs, at Far Edge level. Telco Network Functions will be deployed, in virtualized and/or Cloud Native form, over such infrastructure to enable new services. Telco Cloud interworking with Hyperscale Cloud Providers (HCP) will be considered as well depending on service and commercial needs. To support this evolution towards the Edge, TIM is working with its partners to develop the following technological enablers:

  • a pre-integrated, modular Telco Cloud stack specifically designed for use at the Edge of the network;
  • the automation and orchestration solutions necessary to manage all aspects of the Telco Cloud Infrastructure and the onboarded Network Functions.

In the short-medium term, no mass market services that require latencies/throughputs not compatible with the current TIM network development are expected: development of the Edge in the national sites therefore seems to satisfy almost all applications. The main drivers that can move Edge beyond might be:

  • activation of Ultra Low Latency services and applications (e.g: Extended Reality, Self-driving vehicles, Industrial Automation, eHealth);
  • reliability, security and segregation of data in restricted areas, for services such as video surveillance, Smart Cities, Private Networks;
  • efficient balancing of the computational load between device and network;
  • transport network off loading;
  • decreasing of nodes outages impacts;
  • evolution of fixed and mobile access technologies (ORAN, OLT/LA unbundling).

TIM Brazil also adopts a hierarchical NFVI architecture and the deployment of Telco Edge Nodes is still under discussion. The main candidates to become Telco Edge nodes are the current DCC (Data Center Core) and DCE-R (Edge Data Centers Regionals) in synergy with the dual mode 5G/4G packet core) implementation.

 

Services components and service layer
Regarding the platforms enabling services at the Edge, Telco and Service components will share the same virtualized infrastructure provided by TIM, depending on their requirements. Service components can be provided by several entities: as a private cloud provided by TIM, by Hyperscale Cloud Providers, by “Telco-oriented” emerging technologies (e.g. MobiledgeX, Capgemini Ensconce). Edge computing presents new challenges for application layer, such as managing highly distributed applications and data, as well as orchestrating Edge operations at significant scale or running real-time workloads across distributed infrastructure. With latency being a key issue, applications that require direct connections to centralized services can be risky, so Edge applications might need to combine code and contents into single packages to ensure the best possible performance.
Moreover, with Edge-to-core bandwidth being a significant constraint, it’s important to understand the costs of moving data and building application architectures that are designed to handle distributed data.
We may exploit at the best Edge Computing when a single, complete and comprehensive view of a “global state” is unnecessary, and when this is true, the application becomes much more portable to an Edge platform. The best solutions will often include a combination of Edge and centralized Computing. In such a hybrid computing model, centralized computing will be used for compute-intensive workloads, data aggregation and storage, coordinating operations across geographies, and traditional back-end processing. Edge Computing, on the other hand, can help solve problems at the source, in near real-time. Distributed architectures will enable to place applications at any tier from cloud to Edge, wherever it makes the most sense.

 

On Prem and Micro Edge solutions
Micro Edge concept is defined as an extreme extension of Edge solutions: it allows to share programmability features, orchestration and remote management platforms. Micro Edge is a TIM resource on customer premise, more powerful than a simple network access device because its target functionalities could be dynamically reconfigured over the time on the base of customer needs, avoiding the on prem physical deployment and maintenance of additional dedicated HW devices. Micro Edge solution fits as enabler both for fixed access telco services and for private network services: in this context partnerships with HCPs are also being evaluated, in order to provide a complete solution to the Enterprises that integrates telco functions and cloud functions together.
In the Brazilian market, there is a growing interest in private networks, showing to TIM Brazil that the Edge Computing infrastructure could be associated with this kind of scenario.

 

Federation and Standards
A critical success factor of the Edge deployment is the capability to federate platforms between MNOs. The federation allows customers to experience seamlessly the Edge service across the operator boundaries, creating a cloud continuum, even across different nations. Application providers are hence able to enhance the footprint and reachability of their application over several federated MNOs. In this context, the “Telcooriented” technologies appear to be ahead, implementing the East-West Bound Interfaces defined in the GSMA Operator Platform Group.
Other Standard Developing Organizations and fora supporting the evolution of Edge Computing are 3GPP, CNCF (Linux Foundation) and ONF.

Company Position and Guidelines

The “Beyond Connectivity” concept can be realized only by transforming the network into a real Cloud Native Platform, offering not only pure connectivity but Telco and Service capabilities. The evolution of the network in a Cloud Native environment with Telco Enablers will allow the full network transformation in an Open Telco Service Platform as a cloud-based execution environment for Service Layer offering flexible resources, expandable through the integration of additional modules via clear interfaces (APIs). The platform must be developed on top of a TIM Cloud Native Infrastructure, as this is the reference technology for the IT, Web Services space and industry.
The target is a hierarchy of TIM Edge Nodes from national peering down to the on-premises deployments in an Edge continuum. TIM Egde Node, providing connectivity to all end-customers through public (or even private) coverage, will either locally break out the traffic on an appropriate service component or will allow the non-Edge traffic to flow to centralized cloud instances or the internet, based on the specific application requirements. Among the Telco components to be deployed at the TIM Edge Node the User Planes of the Fixed and the Mobile Core Networks must be included.
A simple list of these elements might be:

  • User Plane Function (UPF), with respect to the 5G core;
  • Disaggregated BNG User Plane (DBNG-UP), with respect to the fixed core;
  • Access Gateway Function User Plane (AGF), with respect to the context of 5G WWC (Wireless Wireline Convergence);
  • Open Cache, with respect to the distribution of web/video contents.

The deployment at the TIM Edge Node of the above User Plane functions, in a distributed fashion, is a key point to provide the network with the flexibility needed to reach the most convenient service components among the multiple ones available in the network. Only the User Plane of the IP Edge functions needs to be distributed, as the Control Plane, in charge of handling the user signaling, can remain centralized: to achieve this, the new Core Network architecture in the context of the Edge Cloud deployment must be based on Control/User Plane Separation (CUPS) functionality. Due to the increase of NFVI sites TIM needs to deploy an E2E Orchestration and Automation solution, being able to handle Telco Enabler (1) service requests, Telco and Service components coordination and lifecycle management, CNI lifecycle management. The scope is reducing human intervention and reduce dramatically service creation (“Everything as Code”) exploiting the DevOps paradigms (Figure 2). Moreover, from a business model perspective, the ownership of the E2E automation platform is strategic to avoid disintermediation from the richest part of the value chain.

cap03-02 E2E Orchestration

Figure 2: E2E Orchestration and Automation Architecture

click here to enlarge the image

Conclusion

The primary business model foresees TIM as the customer facing entity for the Edge service.
The objective is to leverage customer proximity and tight integration of the Telco and Service components to develop appealing Edge computing use cases, in order to remain relevant to the end-user and avoid being disintermediated by Hyperscalers. A commercial and technical process must be defined between TIM and the Service components providers for all the phases needed to commercialize the Edge service (fulfillment, assurance, ...).
Under the assumptions and guidelines described above the Edge Computing represents therefore the opportunity for TIM to gain the role in the value chain by leveraging customer proximity and tight integration of the telco and service components.

Contributing Companies

Contributing companies to this article are: TIM, TIM Brasil, Noovle.

Reference

1) Telco Enablers should be containerized application built upon several simpler components (Microservices) able to expose and use specific API and should be deployed on a specific Execution environment (CaaS or K8S cluster)

Acronyms

AGF: Access Gateway Function

API: Application Programming Interface

AWS: Amazon Web Services

BNG: Broadband Network Gateway

CDN: Content Delivery Network

CNCF: Cloud Native Computing Foundation

CNI: Cloud Native Infrastructure

DCC: Data Center Core

DCE-R: Data Center Edge – Regional

EASDF: Edge Application Server Discovery Function

GSMA: Global System for Mobile Association

HCP: Hyperscale Cloud Provider

IPCEI: Important Projects of Common European Interest

LA: Local Aggregator

MNO: Mobile Network Operator

NFVI: Network Function Virtualization Infrastructure

OLT: Optical Line Termination

ONF: Open Networking Foundation

ORAN: Open Radio Access Network

OTT: Over The Top

PoP: Point of Presence

RTT: Round Trip Time

SD-WAN: Software Defined Wide Area Network

UE: User Equipment

UPF: User Plane Function

VPN: Virtual Private Network

WWC: Wireless Wireline Convergence