ANALYST Q&A: GETTING CIRRUS ABOUT THE CLOUD

Will Townsend

Senior Analyst

Moor Insights and Strategy

Brian Partridge

Research Director 451 Research

Sue Rudd

Director of Service Provider Analysis Strategy Analytics

Deepak Mohan

Research Director for Cloud Infrastructure Services IDC

Top cloud analysts pool their cumulus knowledge to provide insights into the opportunities and challenges on the horizon for operators as they push ahead with cloud deployments.

Q: Where will we see the effects of cloud adoption the most? What is the biggest opportunity it opens for operators?

Will Townsend (WT) Likely on new 5G service delivery in the enterprise. It will bring agility and OpEx savings.

Brain Partridge (BP)

The most outward facing effects will be the speed with which an operator can create and deliver new services. What we won’t outwardly see but will be equally impactful is the potential for cost reductions using less staff, consolidating/closing their own data centres, more automation, reducing CapEx, etc.

Sue Rudd (SR)

Biggest immediate effect of moving to flat request/response (publish/subscribe) software with well-defined interfaces and APIs is the speed with which new features can be added, tested and deployed. Similarly, software upgrades and interoperability testing are dramatically accelerated.

Biggest opportunity for operators is service differentiation and rapid customisation even when competitors use software platforms from the same vendor, since service features and options can be rapidly modified, tested and on-boarded.

Deepak Mohan (DM)

Existing use cases that can see major improvements include logistics, surveillance and monitoring and safety, where cloud based solutions deployed on the carrier edge can replace or enhance existing solutions. New use cases include smart manufacturing, connected cars and augmented reality, where the computing power, latency and bandwidth limitations have historically been hard limitations.

This will create an environment flush with possibilities, and it will be critical for operators to prioritise use cases, and focus on a cohesive set of enablers and partners for their use cases of focus. At the same time, it will be important to pay attention to market demand and recalibrate the prioritisation to reflect demand pattern changes.

Q: What is the primary hurdle to cloud adoption among operators?

WT Tribal knowledge. The traditional infrastructure providers such as Ericsson, Nokia and Huawei all offer telco grade cloud services but AWS and Azure are making investments to compete. Operators will need guidance in choosing the right solution.

Although there has been significant progress at telcos to adopt software-based networking and services, NFV allowed these to be done in a way that mapped easily to the previous physical network nodes or pre-configured vendor packages. This has allowed virtual machines to remain monolithic silos in the network with VNFs that often did not interoperate across vendors.

Cloud service software is based on a computer science paradigm that defines only service functions, including specialised network functions, and that therefore allows a higher level abstraction which is truly separate from any specific physical implementation. This level of abstraction enables the full benefits of virtualisation with decomposed microservices and highly dynamic distributed real time instantiation, but remains a paradigm that many traditional telecoms operator personnel struggle to internalise.

From the perspective of building and operating a cloud platform, a cloud stack is substantially different from the communications stack that operators have years of history operating.

This is a necessary learning curve to traverse, particularly for use cases involving the core network where operators may often prefer to host on an internally operated platform. In scenarios where operators partner with a third party cloud platform provider, typically for the edge network infrastructure or for business services that may be enabled as over-the-top services, operators are abstracted from the details of the underlying stack. Nevertheless, operators will still need to invest in the integration of the cloud services with their broader portfolio, including creating a holistic security posture and providing enterprises/partners the necessary SLA for their use cases.

Building cloud centric skillsets among its internal and external developer community will also be an important enabler for rapid delivery and usage of new services by the operators.

Q: Are there any new challenges or risks that come with cloud-based services and network functions?

WT Yes, lack of knowledge and acumen. Operators have traditionally relied on their core infrastructure providers for single purpose, on-premise hardware. They will need guidance.

Moving to cloud models brings new complexities that require telecom operators to increase their software IQ. This applies to app/dev, test, and importantly runtime orchestration and management across a hybrid, multi-cloud and edge IT environments. Cloudification requires new skills, toolchains, approach to security, governance, etc. Cultural inertia is a real threat that must be managed. Executive management sponsorship and active participation is key.

Many of the new risks for telecoms operators come from blind adoption of the limited capabilities of the IT data centre cloud. For example, Kubernetes automates deployment, scaling, and management of containerised applications, but is not optimised for multiple Kubernetes clusters distributed across multiple network domains. Nor does it address issues such as dynamic microservice load management across distributed containers.

In addition, the inefficiency of application-level service mesh communications - like istio - for end-to-end service assurance is becoming apparent.

The order of magnitude increase in very low latency control plane transactions, either between two service functions or from a service function to network data storage function in the telco cloud, require handling massive numbers of low latency events at response times up to two orders of magnitude faster than any IT data centre.

As with any new technology, there will be a learning curve for operators and partners to gain familiarity and expertise with the cloud platform. Important among these are operational activities like enforcing security and data access policies, and ensuring oversight capability in a cloud native environment. Investing sufficient time and effort on this learning effort will be critical, to avoid failures that may be caused by configuration mistakes or incorrect usage of the platform. Such errors may end up getting exposed as reliability or security issues, setting back the progress on the operators cloud journey. Investing in this learning effort and reinforcement of the process and skillset changes will be critical for operators to build maturity and confidence in operating cloud based service sets.

Q: What are three key considerations operators should weigh when looking to add a cloud element to their network?

WT Cost, ability to monetise new 5G services, and ability to improve capacity and quality of service.

Can I support internal standards and external expectations for experience, security, availability, cost efficiency? Do I have the skills on staff to achieve the outcomes we require, or do I need to hire, train or outsource?

Three fundamental considerations are: usefulness of the cloud element (or elements) involved, demand for early 5G standalone (SA) services, and lower cost of test and operation.

Some cloud elements can profitably be brought in alone into a non-cloud based LTE network. For example, it may be highly beneficial to bring in unified data management early with interfaces to legacy data stores embedded in VMs to begin to separate common subscriber data from vendor specific solutions e.g. for subscriber analytics.

In most cases operators will likely prefer to bring in a complete sub-set of service functions required to deliver a complete service such as network slice selection and management for URLLC Industrial applications, or functions required for voice, messaging and data service that are part of fixed wireless access (FWA).

Cloud elements will rarely be isolated components within operators’ IT portfolios. In order to effectively incorporate cloud elements, operators will need to integrate it with their broader IT assets, and deliver an overall SLA and unified experience to end customers. Critical considerations for operators on this path would be creation of a cloud-centric operating model that integrates the cloud elements to other elements of the operators fleet; enabling unified security, governance and oversight across the integrated footprint; enabling the broader set of technologies, including automation, code delivery, and orchestration, to support scalability of solutions built on the integrated platform; and skillset investments, to effectively operate and deliver services on a cloud based platform.