Cloud-Native Computing Promises Nimbleness, but Can CSPs Seize the Opportunity?

On December 9, 2019, TelecomTV hosted a symposium that aimed to elucidate the term “Cloud-Native” and why it is becoming a key strand for continued and sustained growth in the telecommunications industry. Participation was rich and diverse from both the supply and demand side of telecoms. Communications Service Providers (CSPs) such as Deutsche Telekom, Swisscom, Telefonica, Verizon, and Vodafone were active participants. Amdocs, Ericsson, Nokia, Red Hat, and VMware were some of the vendors that actively engaged in the discourse. TMForum, the European Telecommunications Standards Institute (ETSI), and Telecom Infra Project (TIP) were representatives from standards bodies and industry associations that debated a harmonious coexistence of telecoms standards and open source innovation.

There is, by now, widespread agreement that cloud-native computing is bound to be a key enabler for a more ‘digital’ telco ecosystem. ABI Research stresses this point in our recently published Cloud-Native Computing in 5G Networks (AN-4898) Application Analysis Report. Webscale companies like Alibaba, Facebook, and Google are effectively using cloud-native software to efficiently develop new services. By contrast, telecoms are just beginning to come to grips with the specifics of cloud-native computing, due in part to its rigid technology and organizational setup. What stands to support this assertion is a remark from TM Forum regarding a Request for Proposal (RFP) for a 5G “technology refresh.” It is important to note that the RFP’s contents a year after it was initially issued were drastically different from the initial version. This constitutes evolution. We know structural changes await the industry when it is on the receiving end of changes to the RFP process.

The Cloud Native Computing Foundation (CNCF) houses many cloud-native initiatives and projects including Kubernetes. A peek at the CNCF’s project list quickly establishes the provenance of cloud-native computing—the Internet domain. The Internet field provides best practice for a telecoms industry that, by and large, remains anchored to rigid and imperative (script-based) processes. Cloud-native computing equips CSPs with the ability to obtain commercial utility and take advantage from technologies pioneered by webscalers. From an infrastructure point of view, an apt analogy is that of pets versus cattle. The former constitutes infrastructure uniqueness (i.e., mainframes, load balancers, and firewalls) and is therefore not easily replaceable. The latter promotes immutability, infrastructure that holds no unique properties and is therefore easily replaceable. If one instance fails, another one can take over. Fitting examples are web server arrays and big-data clusters, technologies that are already elevating CSPs’ ability to probe new commercial fronts.

The industry is making great strides toward cloud-native architectures, which enable CSPs to redesign operations and processes to address three broader areas: networks, Business Support Systems (BSS)/Operations Support Systems (OSS), and new digital services. Telco networks remain anchored to a heterogeneous technology ecosystem, not all of which can be mirrored into a cloud-native equivalent. For example, control plane signaling is complex due to a synchronous communication pattern that is provided by proprietary network equipment. BSS/OSS are, in fact, Information Technology (IT) platforms and not as specialized and therefore lend themselves well to more scalable and agile forms. Ultimately, the ability to tear up/down new digital services is the end goal, and one that vendors are pursuing. But today’s telco software brittleness stands in contrast to the new, malleable version coming from the open source world. To drive that point home, there is very little activity in the open source domain for a generic telco-grade load balancer.

Open source and, by extension, cloud technologies afford nimbleness, but whether or not CSPs can seize the opportunity remains to be seen. Telecoms is driven by standard bodies that have long cycle times to next-generation technologies. On the other hand, open source is characterized by a DevOps approach that moves faster. Though CSPs are at different time slots in their digitalization journey, they should collectively propel the cloud-native agenda forward. A close collaboration between standard bodies and open source communities is a step in that direction. The 3rd Generation Partnership Project (3GPP) Release 17, for example, is already aligning with market trends by considering webscale technologies (containers and microservices) in its future releases. To further that aim, the industry should reach an agreed vision by amalgamating the best of breed from standards and open source. An approved vision may not be easy to come by, but the industry must come to terms with what is now a universally accepted axiom; new revenue forays—like 5G enterprise—will fall flat if the industry relies on long standard timelines and excessive tinkering.

Cloud-native computing raises a key question for CSPs: which applications should they transition to cloud-native equivalents first? An effective strategy would be to look for leverage points. For example, CSPs should locate places in the “old” operations, where a cloudified equivalent can have the most impact and trigger a large shift in behavior. A successful transformation warrants that existing network code is re-written using cloud methodologies. While Network Function Virtualization (NFV) did not fully translate to commercial benefits, CSPs must avoid tinkering with it. Instead, they must leverage cloud-native technology available in the market to begin to transition their “value-add” from Average Revenue Per User (ARPU), bits, and subscriber units to enterprise business expertise. AT&T and Telefonica are already utilizing cloud platforms to diversify their revenue streams. Approximately one third of AT&T revenues come from the enterprise sector. The wider CSP community should leverage products available today by pursuing early pilot projects with a Minimum Viable Product (MVP). Commercial benefits will accrue with later iterations once vendors fully operationalize MVPs.

Lastly, a significant aspect to deliberate is the notion of volume operations relative to today’s complex systems. In many respects, these two business architectures are mutually exclusive. A fitting comment from a CSP executive in TelecomTV’s gathering was that “patterns for consumption (in cloud-native era) are quite different.” Though very subtle, what is happening is the increased consumerization of telco technologies. In other words, the purchase pattern of the past was predicated on high switching costs, relatively low volume, high price, and a pay-up-front model. On the other hand, the purchase pattern of the future will be predicated on low switching costs, high volumes, low prices, and a pay-as-you-go model. This newly created nimbleness and speed creates room for innovation and commercial growth for CSPs, particularly in enterprise verticals.

What does an eventual diffusion of cloud-native architectures mean for Ericsson, Huawei, Nokia, and ZTE? It means that, in addition to “masters of complexity,” they need to also become “masters of consumption.” The exact dynamics of that shift will be discussed in a future deliverable. For now, ABI Research posits that the successful vendor of the future will need to shoulder much more, or perhaps all, of the financial risk of CSPs adopting their products and services. Vendors should invest in CSPs’ success and that investment will need to go beyond merely installing and integrating capabilities. They need to guide CSPs’ prospective enterprise customers to consume capabilities they have at their disposal—sometimes even doing it for them.