During a breakout session at its Connect 2017 conference and exhibition, Chinese telecommunications company staged a session designed to examine the concept of what the company has called ‘boundless computing’.
What is boundless computing?
The term itself is meant to describe new approaches to network-level software and systems design. It denotes how a firm like Huawei can position its server line as more than just a box (as hardware)… and something that can support a higher level notion of software application development solutions.
As president of the IT server product line at Huawei, Qui Long prefers to go by the name Red Fox.
Boundless computing mindset
Long urged us to think as follows:
Beyond the CPU boundary – and this means adopting new approaching to heterogeneous computing designed to bring the compute function closer to the data sources… as we move toward use of AI and deep learning in the future, we will start to create computing architectures where different functions happen in blocks i.e. storage+compute or network+compute.
Beyond the server boundary – these are the next generation hardware platforms where literally everything is delivered on demand.
Beyond the datacentre boundary – this part of the theory accomodates for the fact that not all computing will happen in the datacentre and some of it will now happen out on the edge… as in our now understood notion of edge computing.
All forms of boundless computing are intended to optimise and re-optimise server architectures and make traditional (and new) compute workloads more efficient.
Shifting upper layer
To describe the concept again then, boundless computing is all about shifting away from our previous reliance on upper-layer software for resource sharing and architecture orchestration to manage hardware resource pooling.
“Boundless computing includes optimising computing for applications and bringing computing closer to data sources, to unleash the full potential of computing. It also includes pushing beyond the boundary of servers and enabling datacentre-level resource pooling and on-demand provisioning, to boost the overall computing efficiency of data centers. Moreover, it requires going beyond the boundary of data centers, enabling smart access, and taking computing even into the data sources, in this way, to smarten up data at the remote end,” said Huawei’s Long.
With full pooling and intelligent composition of heterogeneous computing, storage and network resources, Huawei claims that its solutions for diverse service requirements and on demand provisioning with benefit from embracing this concept.
The term itself may be slightly proprietary and somewhat internally self-serving and unlikely to become a new industry-wide de facto term, but that’s only a ‘maybe’… this concept of workload intelligence inside more intelligent networks could embed and spiral upwards – that being so, it’s worth having thought about this term.