There are many exciting new applications that are being envisaged or created to take full advantage of what the cloud has to offer. Unfortunately, many of them have requirements that aren’t well supported by today’s networks and network operations.
In order to have the agility and power to take advantage of these opportunities, organizations such as communication service providers and enterprises will need to adopt different approaches to architecting their networks, software design and operations, and to the relationships they have with vendors.
Much is being written about IoT, but these applications, which may result in as many as 1 million sensors or devices per square kilometre, will generate a far higher proportion of control signalling than user data. This is the opposite of services such as video streaming.
Recent versions of the 3GPP 4G standards are looking to address some of these issues, but fundamentally the network architecture and its functions will have to be extremely flexible and scalable to be able to respond if millions of these devices simultaneously come online in response to local events.
There is also a great deal of investment being made in automation. Not just autonomous cars, but, transportation in general, as well as automated industrial processes from infrastructure to manufacturing. These processes will require critical machine communications with latencies around 10ms, and for haptic and tactile use cases, in the region of 1ms.
These stringent specifications will require more distributed cloud-based processing that puts compute resources closer to the automated function. The network will, again, have to be architected differently, moving the cloud closer to the customer or device with ‘edge clouds’.
At the same time, virtual and augmented reality applications, or a combination of both, known as mixed reality, as well as multi-user games that allow participants to manipulate 3D objects, will require both low latencies and ultra-broadband speeds. And, let us not forget, the taste for on-demand multimedia and video continues to grow and there seems to be no end in sight.
The Specifics of a Cloud-Native Architecture
On an architectural level, the only economical way to create the kind of flexibility in network resource allocation that will be needed is to move to a cloud model embracing concepts such as Network Functions Virtualization (NFV) and Software Defined Networking (SDN).
This approach, building on concepts that originated from the data centre, is already being adopted in key areas of the wider network. Although we will operate with a hybrid of physical and virtual network functions for many years, the drive to virtualization will continue. NFV allows for resources to be instantly re-allocated to completely different functions, which will be a prerequisite for modern day networks.
When combined, NFV, SDN and cloud operations have tremendous potential to provide both scalability and flexibility, but only if the underlying software-defined network functions are cloud-native. A cloud-native approach includes network function software disaggregated into smaller components and stateless business logic with common data repositories.
This opens the door for flexible deployment options across the network, with network functions that dynamically interconnect to create whatever services consumers, enterprise users or millions of IoT devices are demanding at that instant. And all of this runs, economically, off a common, universal and adaptive network infrastructure.
Not only will the demands of all these various users, devices and applications require extreme agility in real time, business models will shift along with the rapid adoption of innovative applications as the market responds to the different opportunities.
For organizations, this will require shifting their operations towards greater automation to enable intelligent resource management and allocation, as well as agile development processes and DevOps, requiring organizations and network vendors to work in much closer collaboration.
Collaboration is Key
Agility, contextualization and personalization will be key service differentiators in this new digital world. To achieve this, organizations will need to pivot from network-centric to customer-centric.
As they make this change, costs for network operations will need to be optimized as investment priorities shift to customer-focused enhancements and increased service velocity. Automating the service lifecycle with monitoring, analytics and closed-loop assurance processes will help to optimize operations costs and allow nearly constant feedback loops to developers, making them ‘operations-aware’.
As these changes take hold, a single environment across network and data centre (IT) operations will be formed. IT will not simply deliver applications for operations to deploy and operate, but rather work in collaboration with direct feedback from the operations team during the development process to provide specific insights about network performance and customer requirements.
Reliability will depend on detecting and fixing small defects in the many sub-components of interconnected and service chained network functions early in the development process. Software updates will occur with much higher frequencies.
Vendors will also need to become more involved in the agile development process since they will be responsible for the development of many of the applications and application platforms. This will require a higher level of partnership and collaboration between organizations and vendors than exists today.
For organizations to become cloud-native is a logical evolutionary step and may be seen by many as transformative in terms of business and operations processes. Taking a cloud-native approach is about shifting to a software-based operational environment that is more agile and responsive to end customers and their experiences.
The synergy of the cloud with the network will be key to deliver a platform for society and the economy in this century. To play this role properly will require organizations to be far more flexible and agile to meet all these diverse and rapidly shifting needs.
What are your thoughts on cloud-native architectures? Let us know in the comments.