Posted on

Future Cloud

Introduction

Mark Weiser at Xerox research centre PARC was best-known for his vision of “ubiquitous computing,” a concept he first proposed in 1988.

Mark hoped to create a world in which people interacted with and used computers without thinking about them. Ultimately, computers would “vanish into the background,” weaving “themselves into the fabric of everyday life until they are indistinguishable from it.” With Cloud being everywhere and scaling from simple appliances out to data centres filled with thousands of nodes, we can see this vision becoming reality.

In this discussion we’ll talk about what Cloud 5yrs years from now. And to be sure, at the pace of change we’re seeing today, Cloud will be very different in just 60 months. The best way to predict the future is to look at the past and present trends. This way a possible roadmap forms showing likely outcomes.

Past

In comparison to today, when we look at the past, observations are:

  • Complexity: To drive required business outcomes, applications and data required a complex mix of storage, compute, networking and virtualisation. Complexity resulted in IT Operational silos, Compute Admins, Storage Admins, VMware specialists, Networking Team, etc. And then another team – Project Management because every change has now become a project with many changes becoming a program. Everything is done slowly, and with caution.
  • Move to the Cloud: Us and them culture broke out between Developers and IT Operations. IT Ops resist change because with change comes added risk and yet more complexity – very much a caretaker approach. But Developers are under pressure to innovate. And you can’t innovate on the status-quo. So, given choice, Developers moved to the Cloud. AWS made the most of the shift and appealed to the Developer community with ready-made LAMP stacks.
  • Cloud became trendy: Go back 5yrs years and you see 2 types of CIO – The Caretaker, and The Innovator. Rarely a CIO that could do both. So, the Caretakers want to keep things as-is. But the Innovators are more in-tune with the Developers and need for Business agility. The answer – Cloud First Strategy for Innovator biased CIOs.
  • Infrastructure changes: It’s funny but thinking back just 5yrs infrastructure has changed dramatically. I remember architecting a solution for an MSP required to deliver 1.2PB of usable storage for 2,000 VMs. The storage alone with hundreds of spinning disks occupied over 20 data centre rack units. Now, I can architect a solution that delivers over 2PB of storage with enough compute to host more than 2,000 VMs in just a single data centre rack. The difference – we no longer throw hardware at the problem. The difference is software defined.

Trends

Several trends are shaping Future Cloud.

  • De-complex: Converged Infrastructure (CI) with its many moving parts, and operational overheads has given way to Hyper-Converged Infrastructure (HCI). HCI was worth $1.1B in 2015. It’s now worth $9B and is projected to be worth $14B in 2024. In contrast, CI market share is contracting -1.6% per year, and is projected to be worth only $5.8B in 2024.
  • Move back from Public Cloud: ESG research shows 77% of organisations surveyed have repatriated their mission critical workloads back on-premise. At scale Public Cloud carries high costs. And when transferring many TB of data per day, costs can escalate out of control. Leveraging the Software Defined Data Centre, Smarter Business (and smarter CIOs), a Cloud-Like experience can be given to mission-critical applications close to where the data is generated. We’ll see this trend increase as more data gets generated at the Edge. Gartner predicts 50% of data generated will be at the Edge by 2024.
  • LCaaS: Local Compute as a Service. IaaS in data centres is giving way to LCaaS on customers’ premises. Node-based, simplistic architecture together with software defined storage, compute and networking provides the on-prem Private Cloud with policy-based automation layered on top. Now, turning Cloud on its head, we have the Central Control model. With Central Control, the Private Cloud can be anywhere, even offline for up to 30 days as with MS Azure Stack HCI. So, now we have LCaaS on factory floors working IIoT, we’ve got LCaaS in hospitals taking care of patient records, and even LCaaS on cruise ships and oil rigs.
    • AWS with Outposts, Azure Stack and VMware on Dell EMC are all examples of managed LCaaS. MSPs too are creating portfolio offerings looking to provide centralised control, remote support and management functions to on-premise Clouds. In fact, the MSP market is set to grow by over 11% CAGR reaching $326B by 2025.
    • OPEX models supported by vendors will provide a true private cloud as utility model. For example, Azure Stack subscription charges are based on CPU cores consumed. Into this unit of measure other costs are aggregated, for example disks, memory, server node, software, services, etc.
  • SD-WAN: We can’t think multi-cloud unless we can connect multiple clouds. Enter software defined networking. As the internet and business broadband has increased in bandwidth and more importantly reliability, SD-WAN has become a cost-effective alternative to traditional Telco MPLS WANs. Furthermore, SD-WAN can splice the network and intelligently give dedicated channels to different cloud services. For example, a dedicated channel to MS 365, a dedicated channel to AWS IOT functions, and a dedicated channel to hosted corporate resources. Even a dedicated channel to business partner resources. Say for example a factory system automatically places orders on suppliers for required components, here SD-WAN would provide the secure channel. And Boomi provides the API translation between factory and supplier systems.
  • 5G the Game changer: 5G follows previous mobile technology generations. 3G enabled smartphones, and 4G which enabled faster internet browsing, mobile videos, and the proliferation of connected applications such as mobile banking. 4G also facilitated many new connected services such as UBER and Spotify. However, 5G high-speed connectivity and capacity goes beyond current capabilities. So much so, many studies are trying to predict the effect this enabling technology will have on society.

Paolo Collela writing for Ericsson AB with a long history developing Telco communications technology, reports: ‘5G will enable us to control more devices remotely in applications where real-time network performance is critical, such as remote control of heavy machinery in hazardous environments, thereby improving worker safety, and even remote surgery’ (Colella, n.d). Whereas 4G was very much focused on the smartphone as a connected multi-function device, 5G is seen as the catalyst for everything connected.

According to the Global System for Mobile Communications (2019), IoT connected devices will grow to over 25 billion by 2025. Of these connections, it estimates 3.1 billion devices will use mobile networks. Because 5G supports a massive number of static and mobile devices, future demanding use cases including self-driving cars underpinned by 5G connected smart cities, are now likely. In addition, 5G low latency, ultra-reliable networks will better enable industrial IoT applications. Here, 5G provides increased opportunities for smart factories where sensors continually monitor machines and processes. Then data analysed in real-time by 5G connected Artificial Intelligence will continually optimized manufacturing output.

With speeds reaching 20 gigabits per second, 5G is likely to be 100 times faster than 4G. Another advantage is network slicing. 5G resources can be split to tailor bandwidth, coverage, and security from one network device to another. For example, 5G can enable video streaming while driving a connected car.

Another key difference is latency, (the delay between sending and receiving data). 4G has a latency of 200 milliseconds. 5G has a theoretical latency rate of just 1 millisecond. The ability to send and receive data quickly enables the development of real-time services making 5G the platform for connected devices and machine-to-machine innovation.

But as 5G devices become ubiquitous, security concerns increase as the networked surface of attack grows exponentially. 5G will need to include increased regulations. Network operators and device vendors must implement standards-based solutions to ensure inter-operability that mitigates security risks.

The International Mobile Telecommunications IMT-2020 is a defined international standards specification for 5G. Evaluating IMT-2020 the International Telecommunication Union (ITU) stated in a recent press release: ‘IMT-2020 specifications for the fifth generation of mobile communications (5G) will be the backbone of tomorrow’s digital economy, transforming lives and leading industry and society into the automated and intelligent world’. (ITU 26 November 2020).

In addition to enhanced consumer services, 5G high-speed, low latency and ability to service many devices is small areas will benefit business innovation in ways not possible with 4G.  Industrial IoT now has the connectivity platform on which to develop machine-to-machine collaboration in real-time. Remote operation of machinery from many miles away becomes possible, as does remote surgery where the surgeon uses a console instead of scalpel. 5G will also become the backbone for smart cities enabling future innovations that at present are too demanding for 4G.

As it matures, the benefits of 100 times faster connectivity with 1 millisecond latency will shape business and society far beyond what we’ve achieved already.

  • Micro-Clouds and APIs: We’re starting to see a proliferation of micro-clouds levering simplistic node-based architecture with high-speed connectivity. In fact, edge clouds now satisfy the tenants of IOT equipment, being:
    • Low power
    • Small footprint
    • Remotely managed & supported
    • Provide feedback loop

Micro-clouds dedicated to special functions will become mainstream. For example, smart-cities will have many micro-clouds installed across the whole metropolis gathering data and performing real-time analysis across use cases not yet imagined. And then as we connect functions together using APIs we’ll solution outcomes at business outcomes layer connecting functions together using “solution recipes”.

  • Ultra-consolidation: We’ve been consolidating by virtualising many servers into VMs for almost 20 years now. But increasing consolidation is so much more, for example:
    • Storage consolidation based on larger capacity solid state media, together with intelligent software enables PBs of storage in a single rack. So now just 21 x 15 NVMe drives can deliver over 1PB of high-speed data services. That’s just 2 rack units.
    • Whereas many physical servers can be virtualised hosted within just a few hosts, each VM has a dedicated operating system, and runs a dedicated application – a 1:1 ratio of VM OS to Application. Application containers now offer yet another level of consolidation. For example, many containerized applications running within a single container host (with this host being a VM). No longer a 1:1, but now 1:Many
    • Offloading and sharing functions cross-cloud provides more consolidation.
  • Rise of the appliance: We’re seeing “black-box” solutions based on node-type architectures built for dedicated purposes. HCI solutions lend themselves to appliances. For example, BAE Systems uses VxRail as the platform for battleship command and control systems. As the trend for infrastructure is to get more powerful, but increasingly smaller, Black Box lends itself to more plug-n-play appliances inclusive of connectivity and API plugins.

Future State

Given the past, and current trends, we can see the Future State of cloud will be:

  • Interconnected everything at high-speed: Ubiquitous networking using intelligently switched channels. Smart applications will establish channels on demand, then disconnect.
  • Appliances replace infrastructure: Turn-key, “Black-box” solutions will dominate the Edge, and even the core. And these smart-appliances will be:
    • Low powered
    • Compact, but capable of working PBs of data and provide real-time analytics
    • Include AI will ensure optimal operation (auto-tune)
    • Include many API plugins
    • Be container based, capable of hosting many apps over a shared container host OS
  • Published APIs: APIs will dominate. The ability to connect machine-to-machine and communicate at an application data layer, business-to-business in real-time – that’s game changing. As APIs dominate, we get Function as a Service. Now distributed applications can be compiled cross-cloud. Using a recipe of plugin APIs we get distributed computing close to Mark Weiser’s vision.
  • Pay per use: With Cloud, the model for years has been pay-per-resource (VM, storage, backup, etc.). Next we’ll get pay-per-transaction. These will be micro transactions based on functions. Functions will switch on as called by API, connect, deliver outcome, then pass API call to another function, and so on…

Conclusion

In conclusion, ubiquitous commuting becomes reality. Micro-clouds sitting on street corners running smart city applications, business-to-business, connected functions inclusive of intelligent machine-to-machine communication, and distributed applications compiled of functions as required – that’s the future.

Infrastructure starts to lose relevance as appliances perform tasks previously only capable in data centres the size of football pitches. Cloud now has the bandwidth, compute and intelligence to be everywhere, and inter-connected to everything.

And Cloud will be pay-per-transaction. Micro-transactions will make up cross-cloud applications. It’s not the infrastructure resource that’s important, but it’s utilities everywhere and connected that paramount. Now that’s Future Cloud.