Understanding Cloud Computing: Definitions, Service Models, and Deployment Types

Cloud computing is a model that enables on‑demand, convenient, location‑independent access to a shared pool of configurable computing resources (such as networks, servers, storage, applications, and services). Resources can be rapidly provisioned and released, minimizing management effort and interaction with service providers.

The industrial Internet is arriving, and the next decade is expected to be dominated by cloud‑based business models, prompting many companies to invest heavily in cloud services.

OpenStack’s Newton release, the popularity of Docker, and the rise of DevOps/NoOps have sparked debate: should enterprises adopt the cloud, and if so, which cloud?

What is cloud?

Zhang San: Our company does not store data on personal computers; we store it in the cloud. Li Si: Come on, you have only twenty‑something people, two servers, no virtualization or distributed architecture—can that really be called cloud? Zhang San: Even a cheap NAS is called cloud.

The dialogue illustrates two perspectives: a layperson’s view of cloud as any remote storage, and a technical view that expects virtualization and distributed systems.

The most widely cited definition comes from NIST: cloud computing is a model that enables on‑demand, convenient, location‑independent access to a shared pool of configurable computing resources (e.g., network, server, storage, application, service) that can be rapidly provisioned and released, minimizing management effort and provider interaction. In plain language, it is like electricity—resources are available “plug‑and‑play” whenever needed.

We call this the broad definition of cloud computing . In contrast, the narrow definition focuses on the underlying virtualization and distributed technology stack, often ignoring SaaS and PaaS layers.

Narrow cloud computing tends to over‑emphasize the infrastructure layer while overlooking service models, business models, and the broader societal impact of cloud.

Virtualization and Distributed Systems

Imagine a village: Zhang San’s family has one daughter and excess food (idle resources). Li Si’s family has five sons and a food shortage (resource scarcity). When guests arrive, Zhang’s family shares dishes (virtualization creates a resource pool), and Li’s family borrows chairs from Zhang’s house (distributed systems use the shared pool). Virtualization builds the pool; distribution allocates it.

Virtualization creates logical resources from physical ones (decoupling). Distributed systems enable multiple users to consume those logical resources.

Virtualization includes:

Compute virtualization – one physical machine hosts many virtual machines (full, para‑virtualization, hardware‑assisted, OS‑level, etc.).

Network virtualization – abstracts physical networking into a pool, supporting control‑plane separation, virtual networks, NFV, etc.

Storage virtualization – aggregates storage devices into a logical pool, enabling multi‑virtual‑to‑one mapping, elasticity, and backup.

Public, Private, Hybrid, and Community Clouds

Using a housing analogy: Zhang’s home kitchen is a private cloud (self‑built data center). Li’s restaurant is a public cloud (shared data center). Wang’s private room in the restaurant is a hosted private cloud (or community cloud). When Zhang’s family needs extra capacity, they move workloads to the restaurant (private‑to‑public). Mixing home and restaurant resources represents a hybrid cloud. A community cloud serves a specific group, like a restaurant open only to students.

Private clouds are limited to a single organization and often cannot achieve the full benefits of cloud‑scale resource optimization. Open‑source solutions like OpenStack can lower entry barriers, but operational costs (24/7 management) often make public clouds more economical.

Public clouds serve many tenants, allowing immediate resource reallocation and achieving large‑scale efficiency. Security concerns exist, but for most SMBs the benefits outweigh the risks.

Community clouds sit between public and private, serving a consortium of similar organizations.

Hybrid clouds combine any mix of the above models, offering a pragmatic transition path as enterprises balance legacy private infrastructure with scalable public services.

IaaS, PaaS, and SaaS

IaaS (Infrastructure as a Service) provides fundamental compute, storage, and networking resources. PaaS (Platform as a Service) offers higher‑level development and runtime capabilities beyond the basic infrastructure. SaaS (Software as a Service) delivers complete applications to end users.

Analogy: Zhang sells wheat (IaaS), Li sells flour (PaaS), Wang sells buns (SaaS). Over time, wheat sellers may add flour processing (IaaS → PaaS), and bun makers may offer custom flour (SaaS → PaaS), illustrating the fluid boundaries between service layers.

Storage mapping: block and object storage belong to the IaaS layer, while consumer cloud drives (e.g., personal cloud storage) are SaaS.

IaaS originated with Amazon EC2 and S3, providing on‑demand virtual machines, object storage, and networking.

PaaS evolved from early XAE platforms (custom development environments) to specialized APIs (image recognition, voice, push notifications) and container‑based platforms like Docker.

PaaS essentially supplies capabilities beyond compute, storage, and networking, enabling developers to focus on application logic without managing underlying infrastructure.

SaaS covers email, cloud drives, and most web‑based applications. Modern SaaS emphasizes thin clients, multi‑screen sync, and collaborative features. While SaaS reduces the need for on‑premise software, it also shifts operational risk to the provider.

What Value Does Cloud Computing Bring?

Typical advantages include lower capital expenditure, reduced operational costs, better service quality, elasticity, rapid deployment, and avoidance of hardware procurement. However, high‑performance, highly available instances can be as costly as building a private data center, especially at scale.

The core value lies in societal division of labor: cloud enables large‑scale, fine‑grained resource sharing, driving cost reductions and efficiency gains. Conversely, the same mechanisms can lead to cost capture by providers or monopolistic pricing, similar to a city where most families buy buns from a single shop.

Understanding cloud as an extension of social division helps avoid hype and recognize both its strengths and limitations. From hunting to agriculture, from metal to digital currencies, human progress has always been driven by more efficient division of labor. Cloud computing is the latest manifestation of that trend.

Therefore, resisting or ignoring cloud is futile; embracing it wisely and deploying it pragmatically is the way forward.

Source: Reprinted from iResearch Consulting (©). For infringement concerns, please contact us.