Overview of Cloud Computing: From Physical Servers to Containers and PaaS

Cloud Computing Overview

Cloud computing addresses four core aspects: compute, network, storage, and applications. Compute refers to CPU and memory, network to connectivity, storage to data persistence, and applications to platform services. The first three correspond to IaaS, while applications correspond to PaaS.

Development Stages

Stage 1 – Consolidation: Physical Devices

In the early Internet era, organizations relied on physical servers (Dell, HP, IBM, Lenovo), hardware switches/routers (Cisco, Huawei), and disks/SSDs. While powerful, physical devices suffer from manual operation, resource waste, and poor isolation.

Manual operation: Installing software, configuring switches, adding disks all require on‑site staff.

Resource waste: Small workloads often consume large‑capacity hardware.

Poor isolation: Multiple applications share the same hardware, leading to contention and failure propagation.

Stage 2 – Separation: Virtualization

Virtualization abstracts physical compute, network, and storage into virtual machines, virtual switches, and virtual storage, solving manual ops, resource waste, and isolation issues. Leading solutions include VMware, Xen, KVM, Open vSwitch, LVM, and Ceph.

However, virtualization still requires manual placement of VMs and network configuration, limiting cluster size and increasing operational overhead.

Stage 3 – Consolidation: Cloud Computing

Cloud computing introduces pooling of resources managed by schedulers, reducing the need for manual Excel tracking. Public cloud providers (AWS, Google Cloud) built on Xen/KVM, while private clouds use OpenStack components Nova (compute), Neutron (network), and Cinder (storage).

Private‑cloud vendors typically sell raw resources and emphasize hardware specifications, whereas public‑cloud vendors focus on application‑friendly services such as databases, load balancers, and caches, forming the PaaS layer.

Stage 4 – Separation: Containers

Containers package applications and their runtime environment into standardized, portable units, enabling rapid deployment across development, testing, and production. Tools such as Docker and orchestration platforms (Kubernetes, Mesos) embody this approach.

PaaS Overview

PaaS abstracts application‑level services, offering ready‑to‑use components (databases, caching, big‑data platforms) and automation tools (Puppet, Chef, Ansible, Cloud Foundry) to achieve elastic scaling without manual configuration.

Challenges remain because environment differences (OS version, user accounts, library paths) can break scripts when moving between data centers or cloud providers.

Author: Liu Chao, Chief Cloud Architecture Engineer at NetEase Cloud, with 10 years of experience in cloud computing, OpenStack, Kubernetes, and related open‑source technologies.