What is computing fabric in the network?
Today’s computing continuums are advancing away from remote data centers to the very edge of the network. But could the next computing paradigm lie in the fabric of the networks themselves? We find out below.
The network computing continuum faces a changing reality over the coming years.
As we move gradually into Industry 4.0, the demand for ultra-low latency and high-privacy computing across the network will continue to rise. Complex use cases such as autonomous transport, remote robotics, smart factories and mixed reality will rely increasingly on edge computing solutions to help to ensure time-critical and massive-data execution.
However, as the use cases continue to grow in complexity and criticality, new forms of integrated computing fabric will inevitably be needed to complement today’s edge solutions and bring the rest of the computing continuum closer together.
Below, I explore the prospect of a future unified computing fabric across the world’s networks.
What is computing fabric?
Let’s begin by taking a look at how computing fabric works.
Computing fabric, sometimes called unified computing fabric, has evolved from earlier forms of grid computing.
Essentially, it is an integrated computing infrastructure in which the closely linked computing, processing and network nodes and connections allow it to produce a much more powerful form of computing.
When viewed together, these nodes and connections closely resemble the makeup of a piece of fabric, hence the term: computing fabric.
From a cloud eco-system to a unified computing fabric
Even as we move to more distributed forms of computing, a significant challenge still remains in the fragmentation of today’s cloud ecosystem. When the computing, processing and networking happens across isolated silos – potentially even across long distances – it becomes difficult to satisfy latency, privacy, capacity and programmability demands.
Closing this computing gap – from data center to device – will depend on the collaboration of network service providers and cloud ecosystem actors, such as hyperscale cloud providers, operational technology companies and system integrators.
This collaboration is already embedded inside the networks across the world, evident with today’s LTE use cases. We can make a phone call and access a server on the other side of the world and barely notice how it travels through so many different networks.
As use cases become increasingly time-critical and demand increased privacy, the need for a global connected fabric becomes clear, in which the computing fabric must become even more integrable and seamless. This can be achieved by taking advantage of already existing unified network fabric to integrate computing in. For more than ten years, we have been working on the details on how to bring these pieces together. We call it the network compute fabric.
The network compute fabric
Computing fabric within the network is a form of computing and processing which takes place within the fabric of the network itself.
This can help to satisfy the extremely low latency demands of future use cases, as well as satisfy the computing demands needed to play as a backend for the application. It offers benefits from the perspective of both the user and app developer – allowing industries to deploy applications in a completely different context without the developer having to modify the application.
Let’s take autonomous vehicles as an example. If, for example, the processing power in an autonomous car reaches its limit, there should not be a need for a change in the application as that processing could be offloaded and could happen seamlessly in the backend of the network.
This same logic can also be applied to many more use cases where there is a demand for extremely low latency, massive data, high privacy and also ease of programmability. As an integral part of that, it is distributed cloud computing that is paving the way for the future of industries.
What happens next?
If we focus on why edge computing represents a critical technological step forward, not only for 5G, but also beyond, we can see the new demands of today’s industrial use cases has led us into this new paradigm of computing. Edge computing allows us to reduce latency, while also providing improved flexibility, privacy and efficiency in managing the network.
Future use cases such as the Internet of Senses and autonomous transport systems provide examples as to why edge computing and a fully-integrated network computing chain will be so critical moving forward.
For use cases which demand ultra-low latency and the highest privacy, such as autonomous vehicles and haptic response, only computing which is embedded within the network can provide the necessary real-time translation and processing of the data. It also enables the enterprise to keep the data in-house which increases privacy around the use case.
As the network platform moves deeper into industries, service providers have an opportunity to tailor a service model unique to the enterprise profile, use case and vertical being addressed. At research, we have overcome some of the technical challenges. What now makes this area interesting is the many different variants of business models that are possible. The question is, which role does each actor want to play in each value chain? And which partnerships will they choose to make that happen?
Read Azimeh’s blog post on how distributed compute and storage could improve future networks.