A pc scientist explains why cloud computing infrastructure must change after the AWS crash

The outage started at roughly 6pm AEDT on Monday. It was attributable to a malfunction at one among AWS’ information centres positioned in Northern Virginia in the US. AWS says it has fastened the underlying difficulty however some web customers are nonetheless reporting service disruptions.

This incident highlights the vulnerabilities of relying a lot on cloud computing – or “the cloud” because it’s usually referred to as. However there are methods to mitigate a number of the dangers.

Renting IT infrastructure

Cloud computing is the on-demand supply of various IT sources akin to computing energy, database storage, and purposes over the web. In easy phrases, it’s renting (not proudly owning) your personal IT infrastructure.

Cloud computing got here into prevalence with the dot com increase within the late Nineties, whereby digital tech corporations began to ship software program over the web. As corporations akin to Amazon matured in their very own means to supply what’s referred to as “software program as a service” over the online, they began to supply others the flexibility to hire their digital servers for a price as nicely.

This was a profitable worth proposition. Cloud computing allows a pay-as-you-go mannequin just like a utility invoice, reasonably than the large upfront funding required to buy, function and handle your personal information centre.

In consequence, the most recent statistics counsel greater than 94% of all enterprises use cloud-based companies in some type.

The large 3’s market domination

The worldwide cloud market is dominated by three corporations. AWS holds the most important share (roughly 30%). It’s adopted by Microsoft Azure (about 20%) and Google Cloud Platform (about 13%).

All three service suppliers have had latest outages, considerably impacting digital service platforms. For instance, in 2024, a problem with third-party software program severely impacted Microsoft Azure, inflicting intensive operational failures for companies globally.

Google Cloud Platform additionally skilled a significant outage this yr as a result of an inner misconfiguration.

Profound dangers

The heavy reliance of the worldwide web on just some main suppliers — AWS, Azure, and Google Cloud — creates profound dangers for each companies and on a regular basis customers.

First, this focus varieties a single level of failure. As seen within the newest AWS occasion, a easy configuration error in a single central system can set off a domino impact that immediately paralyses huge segments of the web.

Second, these suppliers usually impose vendor lock-in. Firms discover it prohibitively tough and costly to modify platforms as a result of advanced information architectures and excessively excessive charges charged for transferring giant volumes of knowledge out of the cloud (information egress prices). This successfully traps clients, leaving them hostage to a single vendor’s phrases.

Lastly, the dominance of US-based cloud service suppliers introduces geopolitical and regulatory dangers. Information saved in these large programs is topic to US legal guidelines and authorities calls for, which may complicate compliance with worldwide information sovereignty rules akin to Australia’s Privateness Act.

Moreover, these corporations maintain the facility to censor or limit entry to companies, giving them management over how companies function.

The present finest apply to mitigate these dangers is to undertake a multi-cloud strategy that lets you decentralise. This entails operating essential purposes throughout a number of distributors to eradicate the one level of failure.

This strategy may be complemented by what’s referred to as “edge computing”, whereby information storage and processing is moved away from giant, central information centres, towards smaller, distributed nodes (akin to native servers) that companies can management immediately.

The mix of edge computing and a multi-cloud strategy enhances resilience, improves pace, and helps corporations meet strict information regulatory necessities whereas avoiding dependence on any single entity.

Because the previous saying goes, don’t put your entire eggs in a single basket.

• Jongkil Jay Jeong, Senior Fellow, Faculty of Computing and Data System, The College of Melbourne

This text is republished from The Dialog below a Inventive Commons license. Learn the authentic article.