In mid-2017, computer scientists at Google’s security-research unit Project Zero helped discover two 20-year-old flaws in computer chips that were so widely used it meant just about every computer and network in the world was vulnerable to hacking.
Among those vulnerable to the microprocessor flaws named ‘Meltdown’ and ‘Spectre’ were the providers of cloud computing that offer IT resources from data centres over the internet. Meltdown, for instance, could theoretically allow a hacker who had penetrated a system to access information in a computer’s memory, which is typically shared by many users on the public cloud. By the time the flaws were made public in January 2018, however, the major cloud-service providers such as Amazon, Google and Microsoft had secured their systems and enhanced their reputations for competence and security.
Cloud computing surpasses all expectations
The ability of cloud service providers to remind people of their worth when such a vulnerability emerges is among the reasons why the practice of cloud computing has become a US$176 billion industry over the past 15 years or so.
The practice of renting computing infrastructure over the internet, which is a key segment of cloud computing, is thriving because it is propelled by several favourable forces. One is that rather than having to build an expensive data centre that will be costly to maintain, businesses can access the most advanced computing services immediately and only pay for what they use. Another is that cloud-service providers benefit from such huge economies of scale that they can offer the most-up-to-date services at reasonable prices. By mid-2017, Amazon had cut the price of its ever-improving cloud services at least 62 times in 11 years. And as Project Zero scientists showed, cloud-service providers are better able to secure a system than most businesses or governments.
The sharing of remote IT resources is expected to grow in coming years. Morgan Stanley estimates that by 2021, businesses will send 44% of application workloads to the public cloud, more than double today’s figure of 21%. US consultancy Gartner predicts the global cloud services market will reach US$278 billion by 2021. All up, cloud computing is proving a rare part of technology. It is surpassing predictions of its usefulness and doing so in a relatively trouble-free manner in an era when much technology is shrouded in controversy.
To be sure, any large, fast-growing industry is bound to stir up some issues. Data centres can have technical problems that put them temporarily out of service. Their large energy demands attract criticism, although cloud computing is a more efficient use of resources at a society level and cloud providers are adept at maximising energy efficiency. Issues about ‘data sovereignty’ have prompted some governments to force cloud service providers to contain data within country borders. No part of the internet, including cloud data centres, will ever be fully secure. The European banking regulator worries that the outsourcing of core banking systems to the cloud is a systemic risk. Increasing market dominance by the likes of Amazon and Microsoft could attract the attention of anti-trust regulators.
It must be said too that for all the benefits the cloud offers, not everything will shift to data centres. Cost-wise, for instance, cloud computing may be better suited to variable volumes of data processing. Certain core operational workloads might stay on the premises, as might critical IT and customised processing. The same goes for data that must stay on site for compliance reasons. Another brake on cloud growth is that applications such as autonomous driving still benefit from having computing resources closer to the data source.
But these caveats pale next to the bigger trends. Favourable economics, technological advancements and security concerns will prompt businesses to outsource more IT to the large data centres that comprise the cloud. In the foreseeable future, cloud computing is likely to fulfil upbeat expectations of its benefits without stirring up too much trouble for itself.
Nothing new but Amazon rewrites the rules
In 1996, the term ‘cloud computing’ was coined to describe the concept for people to access software, files and computer grunt over the web. The term seems to have been inspired by drawings where the network was shown as a cloud. (The origins of the term are disputed.)
The practice of sharing resources, however, is far from new. It began when people used terminals with no processing power to access mainframes by ‘time sharing’ – a concept devised in 1955 to overcome the limitation that only one person could use a computer at a time. Sharing of access effectively by timetable for a wildly expensive room-sized computer with ample storage made computing more economical.
While other companies had been offering software in the cloud, such as the sales-automation software provider Salesforce in 1999, years before Amazon Web Services was launched in 2006, Amazon commercialised a new and much broader cloud-computing model on a massive scale.
Amazon’s entry into the industry came about because after the online retailer launched its web-based retail services in 2002, it realised that it only used 10% of the system’s capacity, a common ratio at the time. So, the company asked itself what opportunity the other 90% might provide. In 2018, Amazon’s public cloud-computing business generated nearly US$26 billion in revenue.
As businesses ponder their cloud needs, several forces are helping strengthen the dominance that Amazon, Microsoft and Google hold over the industry outside of China. Among them are the breadth of their global data centres, their technological leadership, and the trust of clients and industry relationships (with software vendors, consultants, etc.) they have built over the years.
To appreciate the size of the networks the cloud providers have built, consider the global cloud infrastructure Amazon has. Over time, the company has built 60 operational zones that help ensure latency and reliability across 20 regions. Each ‘availability zone’ as they are known contains data centres. The most in a zone is 14 while the minimum is two. A data centre, which is typically 7,450 square metres in area, can contain as many as 300,000 servers, with an assortment of servers, storage, networking hardware and proprietary software. To improve reliability and operations, Amazon operates a network that directly connects users to its data centres. This network has about 150 ‘points of presence’ to link with other networks globally, and 89 direct connect locations.
The networks of Amazon and the other providers are likely to grow in coming years as cloud computing keeps proving itself more economical and secure than do-it-yourself, especially in an era of rapid data growth and heightened cyber-security risks and, perhaps, more discoveries of flaws in everyday microchips.
Michael Collins is an Investment Specialist at Magellan Asset Management, a sponsor of Cuffelinks. This article is for general information purposes only, not investment advice.
For more articles and papers from Magellan, please click here.