IBM outlined their vision for telco and edge cloud in a series of informal analyst roundtables in London recently.
Solving a difficult edge problem
“Historically, the industry didn’t really solve the PC management problem, and the edge management problem is 1000 times bigger.” – Rob High, IBM Edge Computing
Rob High focused his talk on the challenge of managing and securing the edge computing – devices and clusters. With potentially 150B edge devices by 2025 this is a major challenge. Edge devices are not heterogeneous, have fixed and limited processing power, and will be provided by large numbers of OEMs. IBM believe that edge native programming will be founded on existing cloud native programming, and particularly containers. Using this foundation, IBM is seeking to put in place the building blocks to solve the edge device management problem at scale, enabling new applications to run efficiently across edge devices and clusters. Example areas where this could be used were described:
- A car today has over 50 CPUs. Whilst some of these will remain closed systems like braking systems, some will in the future become open to application programming, enabling novel applications like driver fatigue monitoring or condition based servicing.
- A manufacturing plant will potentially have machines provided by multiple suppliers. A common framework for edge devices will enable the cost effective integration of these to support flexible manufacturing.
In response to our question, Rob indicated that he expected most of the edge to remain private and, even when managed from public cloud, it will probably be delivered as a virtual private cloud.
Getting edge right in the Vodafone/IBM venture
Michael Valocchi, from IBM Venture and Nitu Kaushal, from Vodafone Business presented on the Vodafone/IBM venture. The venture is targeting medium size enterprises with the combination of IBM cloud and Vodafone connectivity. They publicised a recent win at National Express which was enabling the move of legacy IT to the IBM cloud, and the development of new customer experience applications in National Express vehicles.
“Edge for Vodafone is at the mast” – Nitu Kaushal
Vodafone outlined their progress and vision for edge. They are focusing at private edge at the customer premise (combining MEC and a private network, including LTE and 5G), and distributed edge, providing multi-tenancy compute within the network. Appledore were pleased to see that Vodafone saw distributed edge as being multi-tenanted with network functions. In our existing edge research we believe high levels of multi-tenancy is the key ingredient that will determine the economic success of public edge solutions.
Vodafone believe the Edge will ultimately be at the mast, though it was unclear the current level of progress in achieving this. Appledore’s expectation is of some upfront investment in distributed edge capability, with tactical augmentation of distributed edge in response to customer demand and new use cases.
“The ocean is a very hostile environment” – Don Scott
Don Scott, CTO at MarineAI (part of MSubs), presented on a novel autonomous ship project, part of the Mayflower400 celebrations. Mayflower 2 , a 15 metre trimaran, will travel from Plymouth UK to Plymouth US without any crew later this year. The project highlights, in stark terms, that edge devices, cannot rely on a ubiquitous network connection. Many edge applications will have to rely on intermittent and low bandwidth networks with onboard edge capability.
The project has learnt from the existing work on land based autonomous vehicles and adapted these for distinct maritime needs. The project is using IBM machine learning capability to centrally develop collision avoidance policies, which can then be deployed on IBM’s Operational Decision Manager onboard. Whilst truly autonomous commercial ships are some way in the future, some of the learning from this project can potentially already augment and support a ship’s captain in making decisions on the bridge.