It is widely accepted that using a hybrid- and multi-cloud approach is needed to increase flexibility and reduce the dependency on any single vendor. However, most organizations struggle to operate effectively across clouds, with environments often being disconnected silos with varying primitives, APIs, and management tools. Thus, moving workloads and data between clouds, and implementing organizational security and compliance policies results in substantial friction and associated costs.
The term “Supercloud” is emerging to describe the vision for a new business model that provides a single cloud interface; spanning across a substrate of public clouds as well as on-premises and edge infrastructures. Bringing these various environments under one holistic platform would enable centralized management of workloads and data. This would substantially reduce complexity, streamline processes, accelerate software development, and reduce security and compliance risk.
Digging through the web, one can find a few blog articles and Twitter posts, dating back to 2021, that aim at using the word Supercloud to describe various ideas of how future multi-cloud architectures may look like. They mostly fail at giving the term a distinguishable and meaningful definition. In 2022, it has now gained some more traction. In August, theCUBE hosted a conference called Supercloud 22 (1), with an attempt to discuss the future of cloud computing and what a Supercloud could offer. However, a clear and widely agreed-upon definition of the term is yet to emerge.
This article is a personal insight into the current state of cloud adoption and how the rise of the Supercloud business model may reduce the considerable challenges most organizations face when operating across clouds and their on-premises infrastructure. Furthermore, it will conclude with a proposal for a definition of Supercloud.
Initially, it is important to review the definition of the terms “hyperscaler” and “cloud”, which are frequently used interchangeably, although they refer to two distinctly different concepts.
Hyperscale computation is originally used to describe the architectural ability to scale distributed systems by seamlessly providing on-demand processing power, memory, networking, and storage. Although there are no official criteria for identifying what counts as a hyperscaler, in the past, the data centers of organizations that required to run large distributed sites, such as Google, Facebook, Twitter, and Amazon, were referred to as such. Over time, the data center sizes of Amazon, Google, and Microsoft dwarfed those of other large organizations, resulting in most people only referring to these three as hyperscalers. Cloud computing on the other hand is a business model,
“enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services)” (2)
through a central interface.
The confusion surrounding the terms hyperscaler and cloud likely exists because hyperscalers are currently the leading providers of public clouds, due to the economies of scale that allow them to develop and offer access to cutting-edge cloud services. However, the ability to leverage a cloud business model is not limited by the size of a provider’s data centers.
Furthermore, the cloud business model is also not limited to catering to external customers, since organizations can create, deliver, and capture economic value through an internal cloud offering. The provisioning of IT services through a central interface that is completely controlled and used by a single company is referred to as a private cloud. A platform that offers external access to IT services is referred to as a public cloud.
Drivers and challenges of Cloud Adoption
The wide-scale adoption of public clouds is primarily due to two drivers. The first is the understanding that maintaining and continuously modernizing on-premises data centers comes at an ever-increasing cost. The solution is outsourcing computation to vendors providing managed infrastructure as a service (IaaS). The second driver is the realization that the service-centric hosting of IT platforms and tools occupies expensive IT talent, that could be concentrated on developing and providing business services and processes that give organizations a competitive edge. The solution to this is the adoption of managed platforms (PaaS) and software (SaaS).
However, the outsourcing of infrastructure, platforms, and software to public clouds comes at the very well-known risk of severely reduced bargaining power toward the supplier.
Reducing the dependency on managed platforms and managed software is done by consciously deciding to use open-source solutions. This enables organizations to switch to alternative vendors that potentially offer the same managed software, or fall back upon self-hosting the software.
The infrastructure provides the foundation upon which an organization runs its business processes and services. Consequently, it’s cost and reliability have a substantial impact on an organization’s day-to-day operations and future growth potential. Issues with the uptime and scalability of specific software often have an effect on particular areas of a business, whereas the disruption of infrastructure is often detrimental to the organization at large. This is why being locked in on any single infrastructure provider poses the largest risk. Reducing this dependency is a considerable challenge. In the case of a stressed exit, a lift-and-shift to a different cloud is risky and comes with a substantial cost. This is because IaaS primitives vary extensively across clouds.
This risk of relying on a single IaaS provider can be eliminated by introducing an open source-based platform that can be used consistently across different clouds, which enables workload portability across different managed infrastructures. Using such a cross-cloud homogenized platform layer should be combined with a multi-cloud approach of adopting more than one public cloud in parallel. One can even consider a hybrid-cloud approach, utilizing both a private and at least one public cloud in parallel. This is currently the best way to proactively develop cross-cloud expertise and optimal workload migration abilities.
The rise of Superclouds
The terms hybrid- and multi-cloud can be misleading when used as nouns. A cloud is a single platform that provides access to IT services on top of computational resources, which would imply that a hybrid-cloud provides services that make use of both on-premises and publically available IT resources. The same applies to the use of the term multi-cloud as a noun, which implies the existence of a cloud that sits on top of publically available IT resources from more than one vendor. There have been solutions, such as the open-sourced ManageIQ project, that offered exactly that but were mostly limited to infrastructure services. These are usually classified as Cloud Management Tooling (CMT) or Cloud Management Platforms (CMP). This has become less appealing with the increasing demand for managed platforms and software services, which hide the complexity of the underlying infrastructure and allow for an amplified focus on application delivery.
Although a homogenized platform layer implemented consistently across clouds and on-premises infrastructure is recognized as the de facto enabler for “build once/deploy anywhere”, it is an endeavor that requires considerable effort and investment. Moreover, deploying applications on separate clouds as well as on-premises, even with a uniform underlying platform, results in a somewhat disjointed experience. For this reason, popular platform vendors have started offering a centralized interface for a holistic approach to platform management and workload deployment across environments. This differs from Cloud Management Tooling by focusing on the application build and deployment experience rather than just on the infrastructure or platform layer.
It might be of value to introduce the term SuperPaaS to adequately refer to a service model that provides a centrally controlled managed platform that spans multiple clouds and on-premises infrastructure. At this time it is questionable whether any currently existing solutions can be referred to as such because most add an additional platform and application management layer on top of the existing platform software. This results in a stack of point products that can be combined, rather than providing a single PaaS solution. Nonetheless, it seems like SuperPaaS solutions are only a small step away.
It is intriguing to see that software vendors are progressively expanding their portfolio to provide marketplaces for running out-of-the-box (OOTB) application layer services on their platforms. This horizontal diversification enables an accelerated adoption of platforms and the ROI of its users. Although these platform vendors often develop such services in-house, they increasingly productize open source projects. This is due to the demand for open source solutions in reducing vendor dependencies as discussed earlier. To expand the portfolio of OOTB services available on their platforms, vendors provide access to catalogs of community-provisioned software, bundled and prepared to be deployed on their platforms with a few clicks. This trend is rapidly evolving to the extent that platforms are increasingly transforming into wider solutions that provide a cloud-like breadth of application services.
In the article, The Battle Is For the Customer Interface, Tom Goodwin argues that control of the digital customer interface is what matters most. Emerging companies provide extraordinary customer experience, while only being “indescribably thin layers that sit on top of vast supply systems” (3).
Uber, the world’s largest taxi company, owns no vehicles. […] And Airbnb, the world’s largest accommodation provider, owns no real estate. (3)
Combining the likely soon-to-be-seen SuperPaaS service model with the trend that platforms are rapidly expanding their OOTB application service offerings, software platform vendors may start developing clouds without infrastructure. As described by Jerry Chen in his famous article on Castles in the Cloud,
“the Big 3 [public clouds] have each built up moats of defensibility that essentially guarantee their status for the foreseeable future …[… with] near-infinite resources and unparalleled control of distribution” (4)
Many companies still regard this as a competitive threat, but some realize that they can create value on top of the massive CAPEX of the hyperscaler clouds.
Building clouds that span across and tap into the existing hyperscaler native clouds would substantially reduce complexity, enable the streamlining of processes, accelerate software development, and simplify security and compliance policy enforcement. Moreover, application metadata that specifies operational requirements could be used by intelligent features that control for cost, latency, bandwidth, security, compliance, as well as workload location, and data residency across cloud environments.
This type of cloud will not be limited to tapping into other public clouds but will extend to private on-premises data centers and Edge locations. This way, organizations will gain the ability to truly “build once/deploy anywhere” using one centralized cloud interface, rather than the disjointed and siloed approach to hybrid-cloud that is currently plaguing IT departments.
This new type of cloud begs for the introduction of a new term to distinguish it from the existing hyperscaler clouds. Some are starting to call this new business model Supercloud.
Supercloud can be defined as the next iteration of the cloud business model, providing on-demand access to an ecosystem of platform and application services via one holistic interface that spans as a substrate across public clouds as well as on-premise data centers and Edge devices.
The vision for the emergence of Superclouds is not a mere fantasy. There is a clear and incontrovertible demand for an abstraction layer that enables hybrid- and multi-cloud computing by homogenizing the underlying complexities. Boiled down to its essence, Supercloud computing is a business model that provides this abstraction layer in the form of a superimposed cloud.
It is always hard to look beyond the horizon to predict the future by extrapolating emerging ideas and trends. Before the advent of the motor car, when asked what would improve their journeys, people would have said faster horses. Though current improvements in cloud computing are incremental, they are laying the groundwork for a revolution. Superclouds will be the metaphorical cars that replace horses, increasing speed and reducing manual labor with not just another marginal improvement, but as a new and disruptive approach to cloud computing.
Currently, hyperscaler clouds are general purpose in nature. Looking ahead, it can be expected that emerging Superclouds will provide a similar breadth of services in order to be a viable alternative. However, observing how markets typically react to oligopolies that serve a broad range of use cases, cloud computing will likely be the target of unbundling. New and existing software vendors will likely target chunks of the cloud market share by providing specialized Superclouds. This would allow them to better address the needs of specific verticals by creating a user experience tailored to the unique needs of a particular scope of customers.
Assuming the eventual rise of the Superclouds, there are further predictions to be made on what may follow. One particular concept worth mentioning is the development of cloud infrastructure commodity exchanges. Superclouds would open the door to application-centric cloud brokering, meaning the on-demand scaling of applications to the cheapest available resources across computation environments. This will require capabilities for identifying the cheapest resources at any given time across a multitude of IaaS providers. In theory, this could be baked into the workload orchestration features of Superclouds, however it is more likely that this will giving rise to cloud infrastructure exchange platforms. Superclouds would then connect to these exchanges to request resources, specifying requirements like location and even sustainability criteria. Exchanges may evolve over time to allow the trading of cloud infrastructure resources like any other commodity. This would enable futures trading and hedging, opening the cloud market to greater competitive forces.
What do you think? Are Superclouds the next big thing or just Supercalifragilisticexpialidocious? Feel free to reach out and share your thoughts!