A Changing Telco Cloud Landscape from Core to Edge Networks

Distributed Internet architectures and edge networks have introduced fundamental architectural changes in telcos and associated network stacks. For example, operators’ data centers may increase in number from today’s 100 to potentially thousands; virtual machines proliferate from hundreds to thousands; and networks shift from tens in centralized core environments to hundreds at the “edge.” Further, with compute and networking converging to drive agility and efficiency, today’s cell sites and central offices become infrastructure hubs that facilitate computing at the edge for low latency and highly dedicated use cases. These architectural variances hold potential in terms of both the ability to support new applications and the control over the development of such applications.

With edge networks, intelligence and service logic comes to reside in end-user (edge) devices and hosts (e.g., personal computer clients and servers). This, in essence, means that edge platforms provide a powerful general-purpose network—a platform that supports control and user plane separation and handles signaling information, media content, and a plethora of new services on an end-to-end basis. With a more distributed arrangement, customers can choose different providers for best-of-breed components or can in effect self-provide when an edge ecosystem maturity is in place. But with the telco industry moving the disaggregated network components and stack from the core to the edge and now into radio access networks, there is an implementation risk that rests largely with Communication Service Providers (CSPs). Pre-validated telco-optimized technology blueprints and/or automation capabilities are seen by some vendors as a means to help CSPs understand and mitigate that risk.

Hewlett Packard Enterprise’s (HPE’s) communications group is a newly formed unit that centralizes and unifies its telco capabilities. HPE seeks to offer a joint proposition to customers and a coordinated solution approach to the market. HPE has a complete telco solution predicated on three pillars. The first pillar is the breadth of its portfolio. HPE supports network function virtualization telco cloud with its own virtual network functions and cloud-native network functions, provides business and operations support system products and integration, and supplies an open 5G core stack. The compelling market force behind the acceptance of HPE’s open product portfolio may well be the ongoing effort CSPs to transform their operations. On the other hand, while CSPs can obtain all the benefits of disaggregated stacks, they will need to balance the risk of deploying open solutions. After all, decisions on new technology purchases have always been based on an assessment of risk and reward. To that end, HPE—along with the wider vendor community—should take the lead in aiding CSPs to understand the importance of assessing risk when looking to adopt open, disaggregated platforms.

Second, HPE advocates a “pre-integrated” approach with a set of tools for new telco cloud and edge deployments. Most CSPs do not have the scale and resources to select individual suppliers for core, registration authority, and applications. They seek prepackaged solutions and validated reference configurations that are optimized for specific operational environments. To address that requirement, HPE creates blueprints that achieve extensive testing, documenting, and standard compliance for open 5G core and edge stacks. HPE’s vision to create vendor-neutral infrastructure blueprints is a worthy objective in the market, but this standardized approach may take some time to take off for two reasons. First, the telephony architecture is still highly specialized and tightly controlled and is therefore not conducive to standardized approaches. Second, telco is an industry characterized by tech heterogeneity, embedded vendor solutions, and services that must be refashioned if CSPs are to innovate at the pace of software.

Third, HPE’s portfolio (e.g., 5G core stack) is available on a consumption-based pricing. This takes risk away by linking CSPs’ operational costs to technology take-up. CSPs can roll out faster with potentially less risk attached. On the other hand, it remains to be seen whether cloud-based, software-as-a-service models satisfy the historical—though stable—requirements of telephony architectures. ABI Research finds that this topic is subject to ongoing discussions. There is no doubt that consumption-based products are coming, but they demand new business models for the industry. For example, CSPs are increasingly seeking vendor partners that literally take over in all aspects of technology—from building apps to defining architectures to managing some of their workloads. So, if HPE intends to build and deliver consumption-based technology solutions, they need a services capability that goes beyond its own products to help CSPs de-risk business and initiate a service and operational convergence in a multi-vendor ecosystem.

There are two approaches to de-risk business. The first approach—and similar to HPE’s approach—is to create pre-validated technology blueprints that aim certify components from different vendors. Alternatively, CSPs can manage risk by implementing as much automation as possible. The starting point for the latter must be observability capabilities, for we cannot automate what we cannot see. This involves building clean inventories and clean data pipes on top of existing systems and assets. 5G edge-distributed deployments warrant observability capabilities built in by design and across the board. The ability to log telemetry and manage it in real time to obtain meaningful insight is critical to implement policies and intent-modeled elements. Everything else is an uphill journey, particularly in the early days when CSPs adopted a hybrid cloud model where some network functions are hosted in telco clouds and some in distributed cloud edges.

How should CSPs design a telco cloud that hosts apps assembled from multiple elements that extend from core to edge networks? On that front, the industry should aim to embed multivendor diversity and innovation into telco networks. But there is a cost to integrate multiple vendors, and that remains a challenge for all but Tier 1 CSPs. To alleviate the cost challenge, CSPs seek integration partners that take over and act on their behalf in all aspects of technology. That includes building applications, defining (new) cloud architectures, and managing their workloads. In other words, significant integration is required, and it needs to be industrialized at the product level. HPE’s telco cloud and edge blueprints are a first step in that direction, but the integration capabilities that are needed in a multi-vendor ecosystem can change. Those integration capabilities must be all encompassing and should no longer be tied exclusively to products bearing an individual vendor’s logo.

Lastly, the full potential of 5G, edge computing, and an evolving telco cloud will be realized when they are integrated into the fabric of enterprise processes. Building a 5G network or a telco cloud is a first step. A second step is to figure out how to best integrate the new technology into enterprise verticals’ business processes. That may not be a matter of technology but a matter of the degree of effort required to initiate service convergence and a process change that spans marketing, sales, and so on. At a simple level, the goal of convergence is to seek efficiencies by intelligently combining all the separate service organizational siloes. This arguably is not a one-company effort but one that will almost certainly include technology providers (e.g., HPE, Ericsson, Huawei, Nokia, and ZTE), CSPs, and enterprise verticals. Until that happens, it may be difficult to fully capture the benefit of telco cloud, edge computing, and 5G.