Preparing for 5G: Distributed Cloud Infrastructure

We just published a paper that explores distributed cloud and its current and emerging applications. Please read it!

Video Transcript of Martin Körling at Red Hat Summit

My presentation at the Red Hat Summit was about preparing for 5G and distributed cloud infrastructure. The move to 5G is a bit different than the move to 4G [because] there are many new applications and use cases we see emerging. 4G and the traditional cloud architecture are very much about the consumer – Facebook and YouTube. The move to 5G is much more about industrial applications, business-critical applications, and even society-critical applications.

We believe the infrastructure will also have to change, to what we call distributed cloud infrastructure. It’s a lot about the data flows from the devices, and we will have to move compute and storage closer to the end customer. There are many reasons to make those moves, it’s not only latency. It’s also about regulatory compliance, it’s about resilience, and network scalability. If you think about it, if you have thousands of video cameras in a factory, it’s a good idea to process that video closer to the factory so that you can decrease the data volume, and do machine learning and storage further up in the network.

Then we talk a bit about hardware, our HDS [Hyperscale Datacenter System 8000] and that we have to have different flavors of hardware configurations. HDS is perfect when we’re in a central location or a bit further out, but, as we move further out in the network, we’ll have to package our hardware a bit differently. We will still want to keep a uniform management software layer - that’s what we call the SDI [software defined infrastructure] management layer. We have recently open-sourced an agent that will allow us to have that horizontal software layer across different types of hardware. We’re working in the Open Compute Project on open sourcing that.