Cloud computing use in the Telecom industry has been increasingly adopted during the last decade. It has changed many shapes and architectures since the first phase of NFV that started back in 2012. In today’s data hungry world there is an increasing demand to move Cloud architectures from central clouds to loosely coupled distributed clouds both to make sense from cost perspective by slashing transport cost to anchor all user traffic back to central data centers but also certainly from security perspective where major customer prefers to keep data on premises. Similarly, with the Hyperscale’s and public cloud providers targeting Telco industry it is evident that the future Cloud will be a fully distributed and multi cloud constituted by many on-premise and public cloud offerings.
Since 5G by design is based on Cloud concepts like
- Service based architectures
- Micro services
Hence it is evident that many operators are embarking on a journey to build open and scalable 5G clouds that are capable to handle the future business requirements from both Telco and industry verticals. The purpose of this paper is to highlight the key characteristics of such Clouds and how we must collaborate with rich ecosystem to make 5G a success to achieve industry4.0 targets.
Cloud Native Infrastructure for 5G Core and Edge
Cloud native do not refer to a particular technology but a set of principles that will ensure future Applications are fully de-coupled from the Infrastructure, on atomic level it can a VM or container or may be futuristic serverless and unikernels. As of today, the only community accepted Cloud native standard for 5G and Cloud is an OCI compliant infrastructure. In general cloud native for Telco means a Telecom application as per 3GPP , IETF and related standard that meets criteria of Cloud native principles as shared in this paper, support vision of immutable infrastructure, declarative and native DevSecOps for the whole Infrastructure.
Cloud native is the industry defacto for develop and deliver applications in the Cloud and since 5G by its design is service based and microservice enabled so the basic principle for 5G infrastructure is Cloud native which will support scalability, portability, openness and most importantly flexibility on board a wide variety of Applications.
As per latest industry studies the data in 5G era will quadruple every year this will make Cloud native a necessity to provision automated infrastructures that will be fully automated, support common SDK’s and above all will enable CI/CD across the full application life cycle
Scalability to deploy services in many PoP’s is the other key requirements for 5G along with possibility to build or tear the service on the fly. As 5G deployments will scale so is cloud instances and it is a necessity that future Cloud infrastructure can be scaled and managed automatically
Application portability is the other key characteristics of 5G cloud. As 5G use cases will become mature there is an increasing requirement to deploy different applications in different clouds and to connect them is a loosely based meshes. In addition, as Network capacities and usage will increase the applications must be capable to move across the different clouds
What Cloud means for Telco 5G
Telco operators through their mission critical infrastructure holds a seminal place in the post covid-19 digital economy. Telecom networks use impacts economy, society, commerce and law order directly this is why Telecom networks are designed with higher availability, reliability and performance.
The biggest challenge for Cloud Native Infrastructure for
- Granularity of Telco App decomposition
- Performance acceleration
- O&M and Operational frameworks
Due to the reason that Telco 5G applications need to fulfill special SLA based performance and functions which somehow are not possible in the containerized and Kubernetes based Cloud platforms of today so we must define a Telco definition of Cloud. Similarly, how we will connect workloads E-W is very important. The questions become more prevalent as we move towards edge . The downside is that any deviation from standard Cloud native means we cannot achieve the promised of Scaling, performance and distribution the very purpose for which we have built these platforms ,
Any tweaks on the cloud principles means we can not provision and manage a truly automated Cloud Infrastructure following DevSecOps which is so vital to deliver continuous updates and new software codes in the 5G infrastructure. Lacking such functions means we can not meet fast pace innovation requirements which are necessary for the 5G new use cases specially for the vertical markets
The last and most important factor is leveraging advances from hyper scalers to achieve Cloud and 5G deployments , today we already see a movement in market where a carrier grade Clouds from famous distros like IBM can be deployed on top of public clouds but here top question impacting is whether “abstraction will impact performance” , the one top reason NFV first wave was not such disruptive because we defined so many models and used model to define another model which obviously added to complexity and deployment issues . Cloud Native for 5G Telco need to address and harmonize it as well
Applications for 5G
Application economy is vital for the success of 5G and Edge . However, based on T1 operators’ deployments of open 5G platforms has revealed that just deploying a open Infrastructure is not enough as adherence of Cloud by application vendors will vary and to truly take advantage of Cloud it is vital to define principles for a Infrastructure lead delivery by devising frameworks and tools to test and benchmark the 5G applications classification as Gold , Bronze , Silver with common direction to achieve a fully gold standard applications in the 5G era . Although Cloud native by principles support vision to achieve common, shared and automated infrastructure but it is easier said than done in real practice as achieving a Telco grade conformance for
Telco services is complex that require rigorous validation and testing. Based on real Open 5G cloud deployments and corresponding CNF benchmarking there are still certain gaps in standards that need both standardization and testing.
- Application resources over committing
- Application networking dependence that slows scaling
- Use of SDN in 5G Cloud
- Lack of Open Telemetry which makes customized EMS mandatory
- Hybrid management of VNF and CNF’s
Luckily there are a number of industry initiatives like CNCF Conformance , CNTT RI2 , NFV NOC , OPNFV which fundamentally address this very issues and already we have seen the results . It is vital that 5G Cloud infrastructures are capable to support east to use SDK’s and tools that vendors and developers can use flexibly to offer and deploy different applications in the 5G era.
In Next Part I shall try to elaborate how Open Telemetry and Automation is driving next era of growth using ML and AI driven solutions