Cloud Computing Architecture: Components, Virtualization, and SOA

Posted on Apr 6, 2025 in Computers

Cloud Computing Architecture Components

Cloud Computing Architecture Components

  • Hypervisor: A virtualization layer that allows multiple virtual machines (VMs) to run on a single physical server by managing hardware resources (e.g., Microsoft Hyper-V).
  • Management Software: Tools used to monitor, configure, and automate cloud resources, ensuring efficient operations (e.g., Google Cloud Console).
  • Deployment Software: Enables cloud service deployment, scaling, and orchestration, often using automation (e.g., Terraform).
  • Network: The infrastructure that connects cloud components, ensuring data transfer and communication between users, servers, and storage (e.g., SDN, VPN).
  • Cloud Server: A virtual or physical server in the cloud that processes and delivers computing services, hosting applications and databases (e.g., Google Compute Engine).
  • Cloud Storage: A remote storage system that securely saves and manages data, providing scalability and redundancy (e.g., Google Drive).

Cloud Service Models

Cloud computing offers various service models tailored to different needs:

1. Infrastructure as a Service (IaaS)

Provides virtualized computing resources like servers, storage, and networking. Examples include AWS EC2, Google Compute Engine, and Microsoft Azure VMs.

Pros:

  • Scalable and flexible resources.
  • Pay-as-you-go pricing.
  • Full control over infrastructure.

Cons:

  • Requires technical expertise.
  • Responsibility for managing security and maintenance.

2. Platform as a Service (PaaS)

Provides a development environment with tools for building, testing, and deploying applications. Examples include Google App Engine, AWS Elastic Beanstalk, and Microsoft Azure App Services.

Pros:

  • Simplifies development and deployment.
  • Reduces management overhead.
  • Scales automatically with demand.

Cons:

  • Limited customization and control.
  • Vendor lock-in risk.

3. Software as a Service (SaaS)

Provides fully managed software applications accessible via the internet. Examples include Gmail, Dropbox, Salesforce, and Microsoft 365.

Pros:

  • No installation or maintenance required.
  • Accessible from any device with internet.
  • Cost-effective for businesses.

Cons:

  • Limited customization options.
  • Dependence on internet connectivity.

Service-Oriented Architecture (SOA)

Service-Oriented Architecture (SOA) is a software design approach where applications are built using loosely coupled services that communicate over a network.

Principles of SOA

  • Loose Coupling
  • Interoperability
  • Reusability
  • Scalability
  • Standardized Communication

Components of SOA

  1. Service Provider
  2. Service Consumer
  3. Service Registry
  4. Service Bus (ESB – Enterprise Service Bus)
  5. Service Contract
  6. Service Composition

Pros of SOA

  • Flexibility & Reusability
  • Interoperability
  • Scalability
  • Faster Development
  • Cost-Effective

Cons of SOA

  • Complexity
  • Overhead
  • Security Challenges
  • Implementation Cost

Virtualization Techniques

When the virtual machine software, virtual machine manager (VMM), or hypervisor software is directly installed on the hardware system, this is known as hardware virtualization. Types:

Full Virtualization

Full Virtualization provides a complete simulation of the underlying hardware, allowing unmodified guest operating systems to run as if they were on physical machines. A hypervisor, such as VMware ESXi or KVM, manages the virtual machines by trapping and translating hardware instructions. While this method offers excellent isolation and compatibility, it introduces performance overhead due to instruction translation, making it slightly less efficient than direct hardware execution.

Emulation

Emulation, or full-system emulation, enables a system to mimic different hardware architectures, allowing software built for one type of hardware to run on another. Unlike full virtualization, emulation does not require the host and guest systems to share the same hardware architecture. Tools like QEMU and Bochs allow users to run ARM-based applications on an x86 machine, making it useful for software testing and cross-platform compatibility. However, this process incurs high performance overhead due to the need for instruction translation, making it the slowest form of virtualization.

Paravirtualization

Paravirtualization requires modifications to the guest operating system to optimize communication with the hypervisor. Instead of trapping hardware instructions, the guest OS directly interacts with the hypervisor using hypercalls, significantly reducing overhead and improving performance. Examples include Xen, VMware’s Paravirtual SCSI (PVSCSI) driver, and KVM with para-virtualized drivers (VirtIO). While paravirtualization offers better efficiency than full virtualization, it requires OS modifications, limiting its compatibility with proprietary operating systems.

Software Virtualization

Software virtualization is just like virtualization but able to abstract the software installation procedure and create virtual software installations. Virtualized software is an application that will be “installed” into its own self-contained unit. Example of software virtualization is VMware software, virtual box etc.

OS Virtualization

OS virtualization is a method that allows multiple isolated user-space instances (containers or virtual environments) to run on a single operating system kernel.

Working Mechanism:

In OS virtualization, a single host operating system runs multiple containers or virtual environments using a lightweight virtualization layer. This layer shares the OS kernel among multiple isolated instances, preventing conflicts between applications. Unlike traditional virtualization, which relies on hypervisors to emulate hardware, OS virtualization directly manages resources at the OS level, reducing overhead and improving performance. Eg: Docker, Kubernetes, OpenVZ

Server Virtualization

Server Virtualization: Server virtualization is a technology that allows multiple virtual servers to run on a single physical server by abstracting hardware resources. It improves resource utilization, reduces costs, and simplifies IT management by creating multiple isolated virtual machines (VMs) on one physical machine using a hypervisor.

How Server Virtualization Works:

A hypervisor (also called a Virtual Machine Monitor or VMM) sits between the hardware and the virtual machines. It allocates CPU, memory, storage, and network resources to each VM while keeping them isolated from one another. This allows multiple operating systems (OS) to run independently on the same server.

Types of Server Virtualization:

  • Full Virtualization (Example: VMware ESXi, Microsoft Hyper-V, KVM)
  • Para-Virtualization (Xen, VMware Paravirtualization)
  • OS-Level Virtualization (Docker, LXC, OpenVZ)

Virtualization Implementation at Various Operational Levels:

  • Application level: JVM/NET CLR/Panot
  • Library level: WINE/WABI/LxRun/Visual MainWin/VCUDA
  • Operating system level: Jail/Virtual Environment/Ensim’s VPS/FVM
  • Hardware abstraction layer (HAL) level: VMware/Virtual PC/Denali/Xen/L4/Plex 86/User mode Linux/Cooperative Linux
  • Instruction set architecture (ISA) level: Bochs/QEMU/BIRD/Dynamo

Parallel Computing and MapReduce

Parallel computing is a method in which multiple processors execute different parts of a program simultaneously to speed up computation. It is used in big data processing, scientific simulations, and AI training. Parallelism is achieved using multi-core processors, clusters, and distributed computing frameworks like Hadoop and Spark. The main advantage of parallel computing is faster execution and efficient resource utilization.

MapReduce

MapReduce is a programming model developed by Google for processing large datasets in a distributed environment. It consists of two phases:

Map Phase:

Breaks input data into smaller chunks and processes them in parallel. Each chunk is assigned to a mapper, which processes and produces key-value pairs.

Reduce Phase:

Collects and aggregates key-value pairs from multiple mappers, performing operations like summing, sorting, or filtering.

For example, in a word count application, the map phase counts occurrences of words in chunks of text, and the reduce phase aggregates the results to get the final count.

Applications of MapReduce

  • Search Engines – Google uses it to index and rank web pages.
  • Log Analysis – Helps analyze massive server logs to detect errors.
  • Machine Learning – Used in training models on large datasets.
  • Social Media Analytics – Processes user data for trends and recommendations.
  • ETL (Extract, Transform, Load) – Cleans and transforms raw data for storage in databases.

Parallel Efficiency of MapReduce

  • Dividing workloads evenly across multiple machines (load balancing).
  • Minimizing communication overhead using distributed file systems (e.g., HDFS).
  • Fault tolerance
  • Scalability

MapReduce Infrastructure

  • Hadoop Distributed File System (HDFS) – Stores data across multiple nodes.
  • JobTracker & TaskTracker – Manage and schedule jobs across nodes.
  • Master Node – Coordinates tasks and resource allocation.
  • Worker Nodes – Execute map and reduce tasks.