Why are Telco Cloud Deployments Stalling?

Telcos have been surprisingly quiet about their telco cloud deployments lately, after a flurry of announcements a few years back (e.g., AT&T aiming to virtualize 80% of its network elements by 2020 and Telefónica aiming to virtualize 75% of its network elements by 2020). There have been a few announcements stating cautious optimism about recent developments, but these are nowhere near the optimistic tone of earlier announcements. This is for a reason: AT&T has found out that when virtualization reaches a critical mass, it becomes progressively more difficult, expensive, and complicated to proceed, especially when factoring in OSS transformation, interoperability between physical and virtual network components, multi-vendor support, and many more issues. Other telcos around the world are experiencing the same issues, albeit at a smaller scale compared to the most aggressive leaders in virtualization of AT&T, China Mobile, NTT DOCOMO, and Telefónica.

A consistent problem has been the business case, although NFV and virtualization claim grand benefits that include new services, network slicing, and much more. Both early and current deployments have been focusing on cost savings due to telcos’ cultural inertia, the need to be backwards compatible with the existing legacy network, and overall lack of knowledge and experience with the virtualized domain (on the part of both the telcos and the vendors). 5G will be built on telco cloud concepts and Next-Generation Core (NG Core) will use cloud concepts extensively. However, it may still be 2 to 3 years before NG Core deployments reach large scale, leaving a time window of uncertainty until then. At the same time, certain technical issues are also slowing down telco cloud deployments.

There are two main areas where telco cloud technologies still need to be developed:

• Although NFV and virtualization were both developed using virtual machines (VMs), containers are really more suitable in the telco domain, especially when considering edge computing and micro-services. For example, VMs use shared resources, including the hypervisor and underlying host operating system, which can lead to security and resource issues. On the other hand, the benefits—and commercial readiness—of containers are still not fully understood and further work needs to be done.
• Orchestration is another area where there is competition and rapid technical evolution, with both open source (mainly ONAP) and vendors attempting to address telco requirements. At the moment, it is not clear which option will prevail, but both ONAP and vendor-driven approaches (e.g., Netcracker) are valid options for telco cloud orchestration.

Both of these issues are vital to the large-scale deployment of telco clouds, and while there is significant industrial activity in both areas, neither option in these two categories has reached a critical mass. There are also many other technical and business issues, but the two described above are, by far, the most important technical barriers as of early 2018.

Telcos are not clear about the catalyst for rapid development, either for business or technology. While there are aspirations and end-market potential, there is no clear indication about where new revenue will come from in the next 2 to 3 years. At the same time, telcos are too dependent on their suppliers for innovation, as ABI Research’s recent Telco Innovation Benchmark: Winners and Losers in the Race to Future Revenues (AN-2542) report illustrates. This is in contrast with the web world, where Amazon, Facebook, and Google use whatever technologies are available to them, and if none are available, they build new ones to solve their problems. These companies build platforms to use internally, but also to monetize outside their own organizations, such as Amazon Web Services. Telcos still build their networks using a network appliance approach that addresses a specific domain, rather than consider the long-term viability of each individual investment on the broader network platform.

The development of the telco cloud needs to come from the top-down, rather than bottom-up. Telcos need to understand the market potential and what end markets need, rather than build their current network platforms on a build-it-and-they-will-come basis. This means telcos need to implement end-market services with the technologies they have available to them today: 4G, NB-IoT, existing core networks, early NFV trial networks, and anything that can give them more insight and expertise into these end markets. Only then can they understand what their networks need to evolve to, and design their future network platforms accordingly. Indeed, the future of their business is not in 5G, fully virtualized networks, artificial Intelligence, and big data analytics. It is in today’s reality, but telcos need to risk and grow beyond their comfort zone.