How The Cloud Lowers Barriers to Entry into the Mobile Telco Space

The announcement by Nokia that it is partnering with Google Cloud, Microsoft Azure and Amazon Web Services (AWS) to develop cloud-native 5G services was a significant and inevitable step that all communication equipment manufacturers have to take to remain in the game and meet the ever increasing demand for modern telecommunication systems.

With the rapid drop in cost of compute and storage power due to the economies of scale that cloud services provide, and the inherent ability to dynamically shift workloads across geographical regions and scale both vertically and horizontally with a single mouse click, services that previously required dedicated hardware that took time and money to set-up can now be spun and consumed in a matter of seconds. The shift from traditional DevOps to cloud-first DevOps also means that new and more efficient system architectures that take advantage of the cloud to offer ‘serverless’ and highly scalable computing, which significantly lower the cost of running a service on the cloud while at the same time improving service availability and user experience.

Having worked in the telecom sector for sometime, I have noted the below to be a growing problem:

  1. Telecommunication systems are highly vertically integrated and often deploy proprietary technologies, leaving very little room for interoperability of various vendor systems in one network.
  2. There is a very hard coupling between software and hardware used in telecommunication systems, system upgrades often involve hardware upgrades, making the upgrade complex, time consuming and expensive to carry out

The move by equipment manufacturers such as Nokia and others to develop their systems for the cloud will solve the above problems by them adopting Open-Radio Access Networks (O-RAN) approach on the cloud to create Cloud RAN. This cloudification of the RAN using O-RAN doesn’t necessarily mean that all Cloud-RAN is based on O-RAN but that O-RAN on the cloud can simultaneously solve the two problems I mentioned above in one swoop. Cloud-RAN decouples hardware from software, enabling the adoption of different lifecycles for hardware and software. This will accelerate the pace of innovation in that space and also enable the mixing and matching of various vendors on the same network easily than before. It will also shorten time to market for new technologies and services.

Other than the technical efficiencies that the cloud introduces into the telecom network, adopting the cloud on the telco network also lowers the Total Cost of Ownership (TCO) as it leads to lowered Capex and Opex spend through:

  1. Fast deployment of services: A fully virtualized base station enables both the virtualized distributed unit (DU) and centralized unit (CU) to run on a general purpose processor platform (servers on the cloud). New software and services can be introduced more quickly and more easily because the software code doesn’t need adapting for proprietary hardware. Time savings mean cost savings.
  2. High hardware utilization: Pooled software running centrally on the cloud enables more traffic to be processed on the same data center servers. This more efficient use of processing power allows hardware to be dimensioned according to average traffic loads. The cost of over-dimensioning to meet peak load is simply eliminated, lowering network rollout costs and idle capacity costs.
  3. Multi-tenancy resources.: The same physical servers can host different applications and services. Cost savings are achieved by using spare general-purpose compute resources for applications other than RAN.
  4. Lower operational costs: Centralizing baseband functions in the cloud data center reduces the need for radio site visits, cuts the number of hardware platforms needed, consolidates installation and maintenance processes, and lowers radio site leasing costs.
  5. Greater automation: Open interfaces due to O-RAN allows for integration of artificial intelligence (AI) and machine learning (ML) algorithms, enabling the network to automatically optimize itself to meet unpredictable demand in mere minutes or even seconds.

The biggest and yet unseen benefit however for this cloudification of the base station and mobile network is how equipment manufacturers can now offer mobile networks as a service. The Nokia partnership with Microsoft Azure aims to do exactly that. In this approach, Nokia runs a fully fledged mobile network service on the Azure cloud edge to offer 4G and 5G connectivity to enterprises. The current situation is that large enterprises often depend on a network operator to meet their external connectivity needs while using different technologies for their internal needs. Most enterprises today for example depend on a mobile carriers 4G network to make calls, connect to enterprise resources when outside the office and has to switch to a local Wi-Fi or use the office LAN when in the office. The move by Nokia and Microsoft will enable an enterprise to run 5G in the office (possibly replacing Wi-Fi) and avail the same private mobile network anywhere in the world where their staff find themselves seamlessly. This as you can imagine will significantly lower the time and cost of setting up a mobile network by avoiding high capex initial investment in hardware and paying for this capability as a service on a per-use model. This lowered barrier to entry, means that the running of 4G and 5G networks will no longer be the preserve of deep pocketed telecommunication companies but also enterprises, investors and innovators.

Another benefit is that this move opens up the previously closely guarded and highly proprietary telco equipment innovation ecosystem to everyone and not just the limited R&D teams in those companies. This open architecture and cloud approach is bound to spur new innovations in service and value delivery to consumers. I believe this will also have the effect giving more power to the telco in determining the pace and direction of technology adoption, something that is currently heavily vendor driven. With close to 6/10 of network changes being vendor suggested/initiated, more often than not, the operator is at the mercy of vendors. The move of the mobile network ecosystem to the cloud on an open architecture will change this, to the benefit of the operator and subscribers.

Kenya Needs a Well Coordinated Data Centers Investment Policy and Strategy

The growth of online content consumption fueled by the trinity of cheaper smart phones, social media and 4G is evident everywhere you look. Your average mama mboga, office workmate, spouse is today a regular consumer and producer of content such as video clips, photos (mostly memes) and posts to social media more often. The popularity of Facebook, Instagram, TikTok, WhatsApp, YouTube, video and audio streaming such as Netflix, Showmax and Viusasa says it all.

This therefore means that content providers need to ensure higher levels of service quality by improving their systems to cope with the demand and deliver the expected experience. One example is that these days YouTube Videos rarely buffer like they did 6 years ago because YouTube is storing those popular videos in Nairobi and not in a data center in Europe or the US.

This drive to deliver a good user experience by the providers means that they have to depend more and more on the public cloud infrastructure. This infrastructure is run by cloud providers such as Google, Amazon Web services (AWS), Microsoft, Alibaba, and many more. These cloud providers on the other hand, lease data centers (DCs) from private investors such as the Africa Data Centers and iColo.
The reason why these content providers use the public cloud is because of how it is designed to be fault tolerant and always avail services at the expected quality.

Cloud Infrastructure 101

The public cloud is designed in such a way that cloud services are provided as close to the consumers as possible without compromising service levels and availability. To do this, cloud providers have points of presence where they avail cloud services from. These points of presence are region based, so there would be Asia North region, Asia South, Africa North, Africa South, Europe East etc. Within these regions, they have city regions or zones. For example Africa South region can have Capetown, Durban and Johannesburg zones.
Within these zones they have data centers that are at least 60miles apart from each other and interconnected with high speed cables. To ensure high availability, many cloud providers usually have at least data centers in a zone. so using the example above, there would be three data centers in Durban area separated by at least 60 miles, same for Capetown and Johannesburg

Is Kenya ready?

With the example above, it means that for Kenya to be attractive to cloud operators, we must invest in DCs in a way that will make it attractive for providers seeking reliable infrastructure to provide content from. At the moment, when I take stock of our status, I believe there is room for improvement in as far as preparing the country to being a destination to cloud providers. The location of current DCs and future planned ones doesn’t inspire confidence on the service reliability from these DCs by would be customers.

The Africa Data Center- ADC at Nairobi (Mombasa road) is by far the best run and largest data center in East Africa, closely followed by the iColo data centers in Mombasa and Nairobi (Karen). Safaricom also runs a Data center in Thika town. There is ongoing investment by Huawei and a local partner in a Data center in Mombasa and also the government DC in Konza that is currently in makeshift modular structures as a proper one is being set up.
All these are investments in the right direction, but I think we are dragging our feet as far as investing in world class data center services is concerned. I sometimes imagine what the situation would be if all the Kenya’s empty malls investors had put money into DCs instead? We would become the regional cloud hub for East and Central Africa by virtue of having a more stable economy and political climate and better infrastructure and power supply by far.

Kenya needs to coordinate and catalyze the investment into data centers by private investors for example by giving tax concessions and incentives to anyone investing in a DC at a predetermined location or region. By this I mean that government policy makers can map out how Kenya can coordinate the investment into DCs to ensure that we become attractive to the large cloud players.

This year, Microsoft and Google announced that they intend to be carbon free by the year 2030. By this they mean that their services will run off offices, data centers and networks that rely 100% on renewable energy sources. This of course means that will stop using coal power and opt for greener sources such as solar, wind and geothermal. With Kenya sitting in a region where all these are abundant, there is no reason why Naivasha should not turn from a flower town on the decline into the DC capital of Kenya. Its closer to EA regions than Mombasa, has abundant geothermal and a lake to cool the DCs. Instead, Djibouti is eating Naivasha’s lunch as it’s fast becoming the regional DC go-to city.

The reason I am calling for central coordination on this is because so far, the investments have been driven by other factors other than suitability of the DC locations in relation to other nearby DCs and the investment levels have also been very low. For example, the iColo DC in Karen is too close to the Africa DC on Mombasa road, this doesn’t meet the minimum distance DCs should be separated from each other. A natural disaster or major power outage on Mombasa road would likely affect Karen but not Thika town or Mai Mahiu town where I think the next DC in ‘Nairobi’ zone should be.

With content consumption expected to grow even faster with the adoption of 5G, cloud computing as we know it will also slowly morph into a hybrid of true cloud and edge computing. In the later, the content is hosted very close to the user than before. In the Nairobi case, instead of the content being hosted in a DC on Mombasa road, it would also be hosted in multiple locations near consumers such as at mobile base stations or nearest malls that will offer DC space for Edge computing services.

Please don’t get me wrong, I am not calling for regulation of the DC space in Kenya per-se. I am calling for government incentives akin to the Export Processing Zones (EPZ) concept but on DCs. So as an investor, I stand to gain tax breaks for example if I build my DC at Makindu town to supplement the DCs at Konza. I am however free to chose where I also think I should set it up as long as I am compliant to the existing laws and makes business sense. Saying that Konza Technopolis will answer my call above is missing the point as we cannot have all DCs located in one place. They will not be attractive to customers seeking high reliability and availability of services.