Technologies like AI and VR cannot be fully realised with the cloud capability that is currently available – Sandi Sobahle, Senior Product Portfolio Manager: ITaaS at Internet Solutions
We’ve become so used to talking about The Cloud as if it’s a single amorphous thing in the sky above us - but the future of cloud technology is decidedly multi-cloud.
The closer we get to experiencing all digital services and content in real-time, wherever we are, the closer ‘the cloud’ must get to users and their devices.
Early cloud services were delivered using massive data centres in key regional centres on each continent. This has worked well, up to a point, but increasingly these dispersed data centres cannot meet users’ demands for latency and bandwidth performance.
We are moving towards a global-local approach to cloud infrastructure that offers local service delivery and global reach by adding numerous ‘edge clouds’ to the web of services accessed via the internet.
Edge clouds bring core cloud functions – computing, storing and networking – physically closer to the end-user. This is achieved using micro data centres to process critical data locally before pushing that data to a central cloud.
Latency – a four-letter word
Every minute, 840 people join social networks. Every minute 46,740 images are added to Instagram. More than 4 million hours of content is uploaded to Youtube every day. This – as well as web objects like text and graphics, downloadable media files, documents and so on, applications, and streaming media – is all content that we want immediately, making latency a significant issue for content delivery networks (CDNs).
And it’s not just quantity that we want, it’s also quality, particularly when it comes to streaming media. From 2012, most television and some smartphones offered 1080p FullHD display, driving the demand for high-quality video.
It’s almost impossible to deliver high-quality streams using a centralised delivery architecture, so CDN providers have been forced to expand their footprint by commissioning their own Points of Presence (PoPs) or partnering with local ISPs.
For example, Akamai Technologies, Inc. operates one of the world's largest CDNs, serving between 15% and 30% of all web traffic. The company boasts that thanks to their network of 240,000 servers in over 130 countries and within more than 1,600 networks around the world, a staggering 85% of the world's Internet users are within a single "network hop" of an Akamai CDN server.
The risk of going viral
Whether content – a video, article, or application – goes viral by accident or design, the fact is that doing so brings with it potential risk for the company or service hosting it.
Sudden and unpredictable popularity, without the ability to dynamically scale cloud services required to deliver it, can result in more damage to a company or brand than not offering the content at all.
The answer may well lie in software-defined networking solutions that automate traffic monitoring and elastic service provision and orchestration to create dynamic data centres. This technology is beginning to become mainstream.
Applications at the speed of biology
It might be difficult to imagine a need for an internet much faster than the gigabit speeds promised on 5G networks. But consider the impact of any delay at all on augmented reality applications, or virtual reality gaming, or when using self-driving cars – these are just three examples of use cases on the very near horizon that will require the global-local cloud.
There will be others that require an evolution of cloud services and the networks that link them.
A new cloud service business model
American research and scientific development company Bell Labs, owned by Nokia, suggests that the relationship between global and local cloud providers may follow one of four basic models:
- Local service provider (LSP) offers infrastructure for global cloud services, where the LSP extends multi-tenant infrastructure services to global service providers (GSPs).
- LSP provides hosted ‘as a service’ functions for GSPs, in a Platform as a Service or Software as a Service model.
- Global operators provide local infrastructure, which enables GSPs to offer fully optimised global-local solutions but at prohibitively high costs that will reserve this model for only the largest GSPs considering the most in-demand areas.
- Local operators provide global services through mergers, acquisitions, and partnerships.
Rather than the market being restricted by one of these four models, Bell Labs believes it is likely that a mix of these scenarios are likely, influenced by regional nuances and regulations as well as the business strategies of the respective LSPs and GSPs.
It is interesting that as we move towards a global digital economy where enterprises and their service offerings have global reach, in some ways it is more important than ever to localise delivery. This is consumer-driven, we want our digital experiences to adapt our lives, rather than adjusting our lives to consume digital content.
Already, certain technologies like AI and VR cannot be fully realised with the cloud capability that is currently available. The partnering of local and global cloud providers is inevitable because it is almost impossible for any single provider to offer the end-to-end service that users will demand.
The alternative is an ecosystem of service providers of varying sizes working in concert to offer ‘the clouds’ instead of just ‘the cloud’.