TIM will have a trial release of its ambitious new 5G Digital Business Platform by December this year. The trial is a stage in TIM’s attempt to transform from a telco’s typical linear, service-driven business model into a platform model.
Those keeping an eye on so-called telco transformation strategies should keep an eye on TIM because its strategy is an almost total encapsulation of the network-as-a-platform business model shift, incorporating a 5G access layer, network slicing, AI and analytics led automated operations, cloud native operations and API-based “exposure” of business and network capabilities.
The aim is to be able to expose capabilities of its network to platform partners that use them to develop services and applications for their customers. That means having a network domain that is sliceable, programmable and accessible, with security, IT, OSS and assurance capabilities developed to support that goal.
After initial forays into vendor-led NFV projects, TIM shifted direction to experiment with building blocks of the “digital platform” over the past year and more, with a mix of vendor partner technology and its own internal development.
The platform includes a Cloud Native infrastructure layer, an Intelligence block (big data and algorithmic analytics), plus a sliceable network formed of a cloud native core network infrastructure and a 5G RAN.
It will also adopt, develop and make available number of APIs that expose network capabilities to third parties that can access and develop upon the platform.
The aim is to have a programmable, flexible platform that sets up network slices in response to requests from third parties accessing the platform, via those APIs.
The trial at the end of 2019 is intended to precede commercial deployments some time in 2020. It’s a pretty classical build of a telco transformation, with an end goal of changing the way TIM interacts with customers and makes money. The golden test for TIM, and other industry partners, is whether telcos can build something that customers want to and can engage with in the real world. Can the operator move from slideware, through the lab and into deployments, and will customers access a telco cloud offering that offers hooks into telco assets as they have done with webscale platforms such as GoogleCloud and AWS?
Building the building blocks
Cloud Native and the Core
A key part of the platform will be to have a cloud native operational capability for virtualised network functions. As a first step, CTIO Elisabetta Romano told TMN, the company is currently working on testing of a cloud native 4G-5G core to support 5G in NSA mode. That work has been ongoing all year and includes testing software from Ericssson and Affirmed Networks. Romano expects commercially availability of these products in 2020 – saying that current versions are “not quite commercial but good enough for us to do testing in our lab.”
The cloud native infrastructure layer depends on a cloud infrastructure that TIM is building on RedHat Openstack for NFV and Openshift for containerisation. You can read more about its efforts in these areas here.
Cloudification of the network will eventually extend to the RAN. TIM has previously worked with Altiostar and Ericsson on aspects of vRAN, and it is staying across O-RAN developments as well – something Romano described as opening up the RAN horizontally. However, Romano was not to be drawn on how the operator might look to introduce vRAN and open RAN elements.
“We have not yet, to be honest discussed the deployment and where to deploy. That will follow how we deploy 5G and at the moment we are more following the industrial districts, for industry verticals, enterprise and FWA scenarios. But the point with ORAN will be, ‘What are the first things we want to to open up?’
“So we are horizontally opening up between the DU and CU, as well as vertically by adding software on COTS hardware. These are different dimensions and priorities for us. The possibility to have a remote unit disaggregated from the baseband with eCPRI working is a key starting point. Adding software on COTS hardware is another key starting point – and then a lot of variations can come afterwards.”
Then there is the work on the intelligence block. This involves building out on the data analytics capabilities that TIM already has. “We are already ingesting a lot of information, so we are structuring the way we ingest all the date that is produced by the platform and using this data, developing algorithms in our centre of excellence in Turin that can apply this data in order to generate use cases.” Some of these are more business orientated – such as churn prediction, and some have an operational perspective. For example IM developed its own algorithm to process alarms from the RAN network, and reduced by half the amount of alarms that are manually filtered. “These are examples where we will do radio planning based on all the data from different sites – where customers are, churn prediction and so on. We have already developed use cases and are training algorithms to be applied to the big data infrastructure.”
That intelligence block also feeds data to a key element of the “Communications” block – the network slicing, orchestration and automation capabilities.
“The network is becoming programmable for the first ever time in communications. This is part of this platform. We have the possibility of an application working with our API to to influence the behaviour of the network in a secure way. This closed loop is probably the key distinctive factor coming from the platform and 5G technology.”
Romano added that the distributed nature of the telco cloud means that delivering similar slices at or from the edge – “is also very much a possibility”.
The deployment of an edge infrastructure is another major change within the platform play, and one that Romano thinks can create a different interaction between industrial and other users and the network. The ability to site applications at different points in the network, using a common cloud infrastructure, is another key transformation target.
So the key enabler between the infrastructure and the customer is the exposure layer. Romano said some of this is about API capabilities TIM has already – but adds, “We are working right now to have the more core capabilities that is possible to expose in the network exposure function (5G NEF) – which is a new function in 5G. Right now we are busy building a Proof of Concept regarding exposure, with several capabilities that we can expose so that app developers can come and develop software and put it in the marketplace.”
Romano said that the “key point” of this exposure pillar is that it is not just a matter of exposing northbound, but that the software and app can influence the network southbound.
No standard, but cross-operator co-ordination?
Developing the underlying cloud infrastructure layer has been a case of trial and error for TIM, as it initially went down the NFV path only to find that it couldn’t access the level of operational flexibility and control it had envisaged. But going it alone in cloud-land is also a risk, and Romano is clearly keen to get hooks into other operators’ efforts and compare notes towards a common goal.
In a previous interview with TMN, Romano outlined the operator’s dissatisfaction at vendor NFV implementations, and its requirement for Cloud Native operations. The operator produced a white paper that it shared with a group of T1 operators globally, outlining requirements for its digital platform. Now it has a second white paper nearly ready to print, and says that it includes the results of feedback from other operators.
It’s really very important that operators try to converge as much as we can, even if that’s not a standard
“We are discussing with peers and CTOs in different operators – sharing notes – because every one of us in one way or another is in this transformation. We are trying to be sure that we can converge as much as we can on the use of cloud infrastructure because that will help us to push the vendors to have a solution that works fine with what we use.
“This is a new model for network vendors because you are decoupling, starting with hardware and software decoupling in NFV. Now we are moving one step ahead to cloud native and it’s really very important that operators try to converge as much as we can, even if that’s not a standard. Trying to have that de facto standard will help because in this transformation there is not a standard for everything; we saw that already for NFV.”
Chances of success
What does success look like, for TIM? There are technical successes to be gained by overcoming a lot of challenges. TIM’s launch schedule is on the edge of what vendors can deliver commercially in terms of cloud native telco apps and functions. It needs also to construct viable security and assurance capabilities for these new functions. Its network slicing capability is necessarily nascent. Its automation and orchestration capability of the infrastructure is also emerging.
But quite aside from these technical gains are the business questions. These challenge the I (innovation) letter of Romano’s CTIO title, rather than T. How will operators market themselves as platform partners, where is the engagement of the developer of industry vertical handled in terms of slice management, SLA management, liability and insurance? These, and many others, are open questions that will follow on from whatever technical capabilities the platform can deliver. But for now TIM looks worth watching.