Amazon’s cloud service, Amazon Web Services (AWS), suffered a massive outage last week. The interruption affected a multitude of services like Netflix and Disney’s streaming platforms. The issue related to an application programming interface (API), which is a set of protocols for building and integrating application software, affected Amazon’s own services like the Ring and Prime Video.

The hours-long disruption even hampered access to Slack, Trello and other Software as a Service (SaaS) tools hosted on AWS’ public cloud. These outages aren’t confined to AWS.

In November, Google Workplace, which is hosted on the search giant’s cloud infrastructure went offline for a few hours, affecting access to Gmail, chat and other tools for users in Europe. Earlier this year, an issue at Microsoft’s Azure Active Directory caused authentication problems, affecting a subset of its customers across many services, including Office, Teams, Xbox Live and the Azure Portal.

These outages cripple productivity and communication, and raises the question of sustainability as too many services are now hosted on the cloud, which exposes the vulnerability of the Internet and its concentration in the hands of a few firms.

“Cloud networks resemble feudal kingdoms more than ecosystems,” said Sean O’Brien, lecturer in cybersecurity at Yale Law School. “Like feudalism, the cloud will be replaced by better models.”

The cloud ecosystem

Cloud infrastructure has multiple parts, and each one is linked to another to form a single architecture that supports different digital operations.

Although people perceive a cloud environment to be virtual, the platform requires hardware equipment like switches, routers, storage arrays, backup devices, and servers. These components assist in abstracting a single machine’s resources, like computing, power, storage and memory, through a process called ‘virtualisation’.

This process pools all the resources together in a cloud. Then, using automation software and other tools, businesses access resources they need at any given point of time.

Most firms pool their virtual resources on a public cloud, which are mostly managed by large technology firms. And most public clouds are deployed as a heterogeneous mix of services that provide on-demand availability and accessibility.

A cloud kingdom

But the existing system propagates concentration of network resources in the hands of a few large organisations. The industry has been pushing for more centralisation, instead of focusing on the edges of the Internet that hold incredible distributed computing power, O’Brien noted.

The cloud ecosystem is currently dominated by AWS, Alibaba, Google, and Microsoft. According to research firm Gartner, AWS holds roughly 40% of the $64 billion global cloud infrastructure market, a larger share than Microsoft, Alibaba and Google, combined.

As a growing number of organisations pivot to a digital-first approach, the demand for cloud infrastructure is trending up. The overall cloud spending is predicted to surpass $1.3 trillion by 2025 with a compound annual growth rate (CAGR) of 16.9%, according to the research firm IDC.

During the pandemic, Amazon has grown significantly by reinvesting a huge chunk of that revenue back into its cloud service. The e-commerce firm’s dominance in the retail market and its close ties with governments has netted huge rewards, and all that money feeds the growth of AWS to the detriment of rivals, O’Brien noted.

Rethinking the cloud ecosystem

But this domination could end as the cloud-based feudal systems can be replaced by better operating models using blockchain technology, advancements in cryptography, and clever consensus models for distributed decision-making. “All of the elements exist to rethink currently-centralised applications and move toward a truly decentralised or federated model for services,” O’Brien said. “This change makes cloud computing look foolish if we truly value the availability, integrity, and confidentiality of data.”

Some experts suggest Kubernetes as a way to minimise risk as it isolates and compartmentalises network resources for better security and fault-tolerance. But, while they allow system and network admins to better deploy and manage resources, they do not remove the “problem of hegemony”. “Clever planning goes out the window if AWS has a general outage across AWS, or even in one of its regions or availability zones, and your organisation relies upon those AWS resources for critical applications,” O’Brien said.

Another possible way organisations can prepare for such outages is by taking a step towards interoperability and spreading out risk across multiple cloud platforms. But this could lead to massive inefficiencies, complexities, and task redundancies as the same application will be deployed across different cloud environments. So, the solution lies in the time-tested decentralised peer-to-peer models that have proven successful at moving massive amounts of data quickly and efficiently while remaining largely tolerant to attacks and outages. “Organisations serious about eliminating dependency upon third-party intermediaries like AWS need to rethink their infrastructure and network topology. Our current Internet model relies upon large cloud providers to do the heavy lifting for nearly every application, and this has resulted in predictable, concentrated points of failure,” O’Brien added.