Making business sense of cloud and distributed computing for 5G
Industry momentum is strong around fog computing, mobile edge and distributed cloud. Critical capabilities to unlock the full potential of 5G network platforms. Where ultra-reliable low latency communication is poised to change the world. This post is about providing a foundation for connecting distributed clouds and 5G. Looking at the fundamental drivers shaping this new market place.
Four similar nomenclatures in play
It is clear we will see cloud and distributed computing resources at new locations in the future. Outside the two main anchors today, centralized cloud data centers and premises based private clouds. The new market is shaped between the two extremes. Where four nomenclatures circulate:
- Fog computing—a scaled down version of a centralized cloud, deployed to support one or a few industry verticals.
- Edge computing—distribution of centralized public clouds closer to users for a broad spectrum of use cases.
- Mobile Edge Computing—edge computing capabilities injected into the mobile networks closer to radios than existing Packet Core sites.
- Distributed cloud—combining the needs for public clouds and network resources. With a globally adopted standard.
In this blogpost we will focus on the market dynamics around distributed clouds.
Why will workloads be disrupted and moved to distributed clouds
A starting point for understanding market dynamics is to look at which workloads we can expect to move around.
Tied to mobile network platforms we see two movements. Performance critical parts of the mobile core network functions can benefit from being distributed closer to end-customers. And a movement in the opposite direction where virtualized radio network functions are centralized closer to the core network. Where both these movements depend on a distributed cloud.
The location of public centralized clouds is financially driven. Dominated by strategies to minimizer cost in any way possible. Where physical locations are defined by access to a) land / real estate b) power c) climate with low cooling needs and d) powerful transmission to Internet exchange points or mobile network core sites. All fine until customer experience phenomena such as “Alexa, can I get a faster response?” and productivity losses for cloud-based software tools come into play. Performance critical public cloud workloads will move closer to end-customers to eliminate roadblocks. With an expected increase in what we label performance restricted.
The third category is workloads moving out from customer premises and into a distributed cloud. Driven by an inefficient use of resources when based on premises. But not further away from the end-customer than what can be done in a performance neutral way.
Beyond these three existing workloads a fourth category with completely new workloads. Where neither cost for premises based nor performance for centralized clouds are acceptable. These represent a paradigm shift where the relationships between device, network connectivity and cloud applications are changed.
What type of new use cases can we expect to see
The million-dollar question is about defining the use case that will take off first. With low odds for bets with low end-customer adoption barriers, high growth potential, high performance needs and clear monetization potential.
First, use cases with high amounts of data in play. Either raw data generated close to end-customers, like in a connected car. Needing processing before distribution to the location of data interpretation. Or compressed data requiring compute resources close to end customers, as for video decoding or augmented reality tasks. In both cases for data flowing to end-customers. Both these use case types benefit from compute in distributed clouds. To minimize transmission of bits with a very low level of information content.
Second, use cases where new performance levels disrupt the existing market preference. Networked gamers are today restricted to use wired connectivity in general and restricted to local area connectivity. Distributed clouds and ultra-reliable low latency connectivity can open-up the market for wireless and wide-area connected gamers. See this as an example of use cases requiring low latency to be considered at all.
Third use cases were compute resources, and the relation between on device and cloud-based computing can be disrupted. Where costly on-device compute, and associated power consumption, can be moved from devices to a nearby distributed cloud. For any stationary device without a strong natural connection to wired power and connectivity. Where distributed clouds provide significant leaps down in device cost and increase in battery life times.
The fourth use cases are about compute tied to customer experience management. Where large amounts of data are collected about products and services in use. With the purpose to either improve offerings or to evolve business models for how the services are offered. These uses cases will leverage artificial intelligence and machine learning in close to real-time for to create step-shifts in customer experience. Distributed clouds enable the execution of these compute and connectivity process to occur faster than humans can notice a difference.
Who will deploy distributed clouds
The eco-system around distributed cloud is complex with a larger number of stakeholders wanting to participate in this new opportunity.
- Mobile network operators—building from critical low latency 5G connectivity services and network slices optimized for performance critical use cases.
- Public cloud providers—bringing parts of their proven cloud services offerings closer to the 5G action.
- Real estate owners—with wide spread under-utilized urban assets that can be repurposed to host distributed clouds.
- Utility providers—offering power, dark fiber and transmission services.
- Customer experience data owners—with an interest to find optimal distributed cloud hosts to secure their customer experience evolution.
- Technology providers—building from the solution innovations enabling the distributed cloud disruption to take place.
An area representing significant cross-industry eco-system collaborations. Where we can expect business models shifts, as the market is shaped. With existing roles, relationships and models in the eco-system changing. Where eco-system stake holders initially focus both on maximizing the size of the distributed cloud pie AND getting a share of this exciting cake.
What can be predicted about the future
All stakeholders will seek to deepen their understanding and to stake out positions the coming years. Where a few things stand out as solid predictions:
- The new network platforms based on 5G, network slicing and distributed clouds have far reaching disruptive implications for a large number and variety of industry stakeholders.
- There is a significant job remaining to maximize the size of the distributed cloud pie, in a way to maximize the growth potential.
- A larger variety of innovation bets will be placed along the journey, driving market creation in many small increments with pivots along the road.
Thank you for investing time and reading to the end. Ericsson look forward to engage with you, and your company. We hope you find this blog post, and related podcast, exciting as a base for deeper discussions. About how you can connect your 5G strategy to distributed clouds in a way that make business sense.