**1. Introduction**

MIT professors J.L. Hennessy and D.A. Patterson in their 2018 Turing Award lecture gave an excellent overview of the history of computer architecture and the lessons of this history in Ref. [1]. In the conclusion of this paper, they wrote: "The next decade will see a Cambrian explosion of novel computer architectures, meaning exciting times for computer architects in academia and in industry." This "explosion" brings great opportunities for computational infrastructure. To paraphrase the title of their paper, we can say that we are going through "A New Golden Age for Computational Infrastructure." The term computational infrastructure will be treated here as the architecture of computational infrastructure.<sup>1</sup> Organization of computations is one of the pillars of human civilization. Therefore, it is important to understand the main trends and prospects of its development, to understand what problems will need to be solved.

<sup>1</sup> This paper is extended version of my talk on the MoNeTec-2022 published in [2].

If we have a look at the history of the progress of computing infrastructure from the individual use of the first computers, the packaged organized computation on stand-alone computers to the client-server paradigm of computation based on high-speed data communication networks and large-scale (giant-like) data centers (DC), then the main lesson of this history is that the main drivers of this progress were the requirements of applications. Nowadays, the computational infrastructure paradigm is moving from "build your own" to the new one—"consume as a service" where a business does not need to buy and develop its own computational infrastructure, rent channels that connect it to the public network, hire expensive professionals for system and network administration, and so on. In the new paradigm, one can request and get the resources and services they need based on the model of "pay as you go." Also, computing paradigm based on the giant-like DC is being replaced by a new one—small cloud edges in [3]. Our applications became more and more real-time applications. So, the time for communication between user terminal and large scale DC where our ability to control communication delays became more and more critical for application operation. The increased restrictions on the interaction time between the application and the terminal device led to a contradiction with the concept of computing based on large-scale DC. Carbon footprint of such organized computation is a heavy burden on our ecology.

The necessity of this change has come from the requirements of new applications with their real-time interactivity, video streaming, and 5G communication. Over the past 10 years, cloud computing as the computing paradigm has completely changed the landscape of computational infrastructure; for example, see [4–7]. In the paper [3], I wrote: "It significantly contributed to the growth of both the number of data processing centers (DCs) and their size, the increase in throughput capacity of backbone channels [5], the increase in equipment density: virtualization of IT equipment in cloud architectures allowed to fit into one rack what previously required 10 racks improving and developing the capabilities of personal gadgets, various types and uses of sensors, the development of data transmission technologies such as OTN, 5G networks, network convergence, the emergence of SDN and NFV technologies gave impetus to the development of a large number of real-time applications (RT applications) in Ref. [8, 9]. Here are just some examples of such applications: smart city, smart home, healthcare (especially its areas such as surgery, telemedicine, emergency cardiology), interactive games, training, augmented reality, agriculture, infrastructure for scientific multidisciplinary research in [7], social communications, energy management systems (smart grid), wireless sensors embedded in a variety of robots, monitoring and control of transportation systems and facilities, assembly lines and production lines, gas and oil pipelines.

An important aspect of the computing infrastructure is power consumption. According to 451 Research, a technology research group within S&P Global Market Intelligence that provides a holistic view of innovation across the entire enterprise IT landscape, the computing power as well as the engineering equipment of all DC worldwide is estimated in 2022 about 200 GW in [10]. This means that in 2022 the energy consumption of all DC was 200 GWt 24 hours 365 days = 1 752 10<sup>12</sup> Wt hours! Thus, the carbon footprint of the contribution of computational infrastructure was about 1.8 10<sup>15</sup> Wt/year (53% of the US). This means that the organization of the computing infrastructure has a significant impact on the ecology of our environment.

*Network Powered by Computing: Next Generation of Computational Infrastructure DOI: http://dx.doi.org/10.5772/intechopen.110178*
