The physical reality of the digital world

In his novel The Name of the Rose, Umberto Eco describes the organization, data storage and access to knowledge in an abbey library seven centuries ago, which eventually burned down, destroying the work of the copyist monks, the first artisans of redundancy before the invention of printing and then computers.

Michel Robert, University of Montpellier

This type of event has been repeated down the ages, each time resulting in a loss of knowledge. The most recent episode in our age of dematerialization was the spectacular fire in Strasbourg on March 10, 2021 at a data center - i.e. a data storage and processing facility - which had major consequences for users. This incident revealed the fragility of digital infrastructures (computers, servers, storage racks, communications networks, power supplies, air conditioning, etc.), leading to sometimes irreversible data loss and interrupted services (information systems, IT applications, messaging, websites, etc.).

An industrial accident of this kind calls into question the physical reality of the digital world. Over the past forty years or so, this world has been built around computing machines concentrated in networks of data centers that make up the digital cloud. This cloud, which appears to be immaterial, is in fact based on infrastructures distributed and interconnected on a global scale. To date, there are almost 5,000 dedicated or shared data centers in 127 countries around the world, some capable of hosting tens of thousands of servers.

The history of IT is made up of alternations between local and global centralized and distributed infrastructures: centralized in the last century around a single computer, then distributed with the advent of nomadic computing (PCs, tablets, smartphones, connected objects...), then mixed today with services increasingly outsourced to specialized companies (GAFAM for example) to save and process data, or communicate via social networks, or in telecommuting via videoconferencing and shared documents.

How do you store data securely today?

If we look at individual usage, 30 years ago data was stored on diskettes with a capacity of 1 megabyte (¹⁰⁶ bytes), then CDs, USB sticks... Today, a personal magnetic hard disk of 1 terabyte (1,012 bytes, i.e. a thousand billion) - the size of a smartphone - represents the equivalent of a million diskettes at a cost of a few dozen euros. Backing up data permanently has been a matter of course since the early days of computing, with unreliable hardware and software at the outset.

Today, we're witnessing an explosion in the amount of data linked to our uses, such as instant backup in the cloud of photos and videos captured on a smartphone. It's also a time of all the known forms of hacking and cybercrime. Safeguarding data requires precautions, such as secure storage facilities.

At the professional level, many users and companies are unable to afford an autonomous, robust IT infrastructure, given the costs of acquisition, maintenance, security, updating and the associated human resources. As a result, they turn to specialist companies who sell their expertise in data security, whether in terms of confidentiality, protection of know-how or privacy. State sovereignty over access to data is also a crucial issue. The distribution of data and data processing on a global scale - and one day in space, with clusters of satellites establishing communications between servers - offers many advantages, provided that the physical limits of the infrastructures used are well understood, particularly in the event of an accident.

A critical look at current offerings is therefore essential: where is my data stored? How is it protected, secured and backed up? What is the carbon footprint of my digital uses?

How much will it cost to virtualize our IT systems?

Some operators offer turnkey services to meet these requirements. Others offer lower-cost access to machines, leaving customers free to make their own choices, for example when it comes to managing backups - contracts between the parties govern the details of these uses. The notion of service quality is therefore essential.

Good communication on the technologies used and their limits, which are sometimes elusive for users, is essential: what levels of data protection are included in my contract? How often are backups made, and how? In particular, the CNIL reminds us of our obligations in terms of notification, in the event of unavailability or, in the worst-case scenario, destruction of personal data stored in a data center.

The physical reality of the digital world also raises the question of the energy resources required for these infrastructures and for our most energy-intensive uses (streaming video, managing virtual currency, bitcoins). The environmental footprint of our connected digital devices and communications, computing and storage infrastructures cannot be ignored: the global share of "digital" in greenhouse gas emissions is increasing every year and will soon exceed 5%, with energy consumption of 2,000 terawatt-hours, or 10% of global electricity demand.

Technical solutions

Scientific and technological solutions are emerging to make the digital and energy transitions, which are inextricably linked, more reliable and easier to support.

This could, for example, lead to "digital short circuits" for reliability and back-up, associated with each data center and making the most of older generations of computing machines running exclusively on green energy. These machines could be distributed throughout a region, thus limiting the impact of an industrial accident on a given site, by playing on the redundancy of machines to ensure backups.

Indeed, in the event of an accident, a service based on redundant computing resources is always far better than the irreversible loss of digital data, whether for private or professional use. For many applications that do not require high-performance computing, or for the local management of data and services offered to users on the scale of a territory or a smart cityIt is conceivable to couple data production and storage with green energies in a "digital short circuit", whether in terms of energy production or storage. heat recovery or production ofelectricity from renewable sources.

Michel Robert, Professor of Microelectronics, University of Montpellier

This article is republished from The Conversation under a Creative Commons license. Read theoriginal article.