What the telco industry needs from cloud computing

Like everyone else, telecommunication service providers want to benefit from a cloud-native world. But in many ways, the IT industry (where virtualization and the cloud were born) is more tolerant than the telco industry. While service providers seek a cloud-native environment, they often exercise caution in order to maximize benefits from the transition. 

In this series of blogs, my fellow Red Hatters and I explore the adventurous paths and requirements service providers continue to traverse. Like most adventures, we start with a warning: the blog is riddled with generalizations, but we hope any exaggerations will help you understand real-world trends.

The IT journey

The IT journey doesn't have the luxury of a paved road. It's filled with twists and turns, and sometimes when you believe you've almost reached your destination, you find yourself back at the beginning. Enterprise IT keeps you moving, hopping from one release to the next, patching fixes on the fly. It's truly Continuous Integration / Continuous Development (CI/CD).

The telco journey is not necessarily a linear one. While service providers must have one foot in the new world and the other in the old world, finding the balance between private and public cloud, virtualization, and cloud-native will be a never-ending challenge.

Architectural requirements

In everyday life, a home is tailored to its residents. The home of a senior might have doors wide enough for mobility aids, a home for a single person might be an open space, and a family home might consider child safety.

The same is true for the networks our architects build. The design of a network is driven by the needs of the customer and the applications being served.

For example, the telco world strives to fully untether itself from hardware. However, considerations such as performance, dimensioning and latency may drive a telco to a specific underlying hardware. Moreover, regulations, such as lead content, may drive hardware decisions. Thus, software and hardware do remain coupled (as ironic as this sounds for a cloud!), and hardware will be tailored to telecom workloads, even in a cloud world.

The architectural design of the telco cloud is driven by compliance with the principles of the tiered data center approach. Many of the same principles that guided the traditional data center continue to guide the cloud world: separation of workloads from data, security principles and networking guidelines. Tenant isolation must be enforced by a network policy, both as a security measure and to prevent resource cannibalization by competing tenants.

Automation, along with ease of operability and serviceability, are central to cloud architectural requirements. A challenge of the legacy world was that each network element brought its own operation, administration, and management (OAM) implementation, so the service provider had to understand a plethora of tools and procedures. 

While addressing OAM consolidation, service providers need to move to holistic, automated operations with a zero touch approach. Service providers usually prefer to prevent human errors by means of automated processes for all life cycle management (LCM) activities, ranging from deployment, configuration, and security hardening all the way through to system upgrades.

Related to the above is the need for centralized fault and performance management. Ideally, this offers a single source for all monitoring needs across the network, empowering service providers with a single view of the health of their operations. 

Metrics should not only provide information about resource usage so log information for ongoing and future analysis. 

As a home to microservices, the cloud platform must be able to serve state-less and data-less entities. This is especially important for life-cycle management (LCM), including upgrade and scale-out activities, where the ephemeral nature of data ensures ease of mobility. 

Choices involving where and how data is stored may change as a result of the move to cloud. While local storage remains an option (along with its challenges), some service providers prefer using storage nodes or external storage solutions. High availability and redundancy must be considered too, while not neglecting the associated costs. 

Indeed, a redundancy model should be used at all levels of the system, contributing to the resilience and high availability of the cloud platform. In this way, failure of some parts of the system don't result in an overall loss of service. Service providers expect redundancy at the control plane, at the networking level, and in storage (including a recovery solution when data is corrupted or as a result of disaster). 

A well-designed network architecture contributes to improved performance and can also address security concerns. Network separation should be done between the service provider and the outside world. This can be achieved by network policies defined by the service provider. 

Security is a byword for telcos, driven by national security standards and technological trends. An automated hardening process must address all aspects, ranging from networking, node hardening, centralized (and automated) user management, and certificate management. 

Traffic from user endpoints should be processed by a firewall capable of preventing Denial of Service and other attacks by malicious users. 

The hardware aspects of the solution cannot be ignored. While acknowledging that the move to a cloud solution empowers the service provider to untether a solution from its underlying hardware, thought must be given to which hardware is chosen in order to address basic needs, such as performance. 

Conceptualizing a comprehensive solution drives hardware requirements and costs. There's a link between how a system is architected, what hardware it requires, which cloud platform it runs on, and what applications it serves. 

Finally, a solution cannot be viewed as a composite of all these parts. Instead, it must be designed and tested holistically. The whole is greater than the sum of its parts!

This article has described some of the architectural requirements for a telco cloud. In upcoming articles, we'll delve into specifics of some of the topics we have covered above.