8. scalability

Scalability Overview

• Scalability is central to managing increasing web demands and complexities.

• Focuses on the subset of the internet utilizing the HTTP protocol, known as the web.

Scaling the Web (Page 2)

• The evolution of the modern web revolves around scalability.

• Scalability addresses the ability to handle increased demands as complexity grows.

Web Applications (Page 3)

• Web applications categorized as:

  • Traditional websites accessed via a browser.

  • Mobile applications using an HTTP backend.

  • M2M (machine-to-machine) applications where back-end systems communicate over web APIs.

Web Application Architecture (Page 4)

• Utilizes a model-view-controller (MVC) architecture:

  • Model: Back-end data.

  • Controller: Application logic connecting model and view.

  • View: Graphical User Interface (GUI) for user interactions.

• M2M applications do not include a GUI.

Data Transfer over HTTP (Page 5)

• Common techniques for data transfer include:

  • HTML: Hypertext markup language generated by back-end servers.

  • SOAP: Simple Object Access Protocol, an XML-based communication protocol.

  • JSON: Lightweight text-based notation for data exchange.

  • REST: Architectural style combining request method & URI, often using JSON or SOAP for responses.

The Scale Cube (Page 6)

• Web applications can scale in three primary ways indicated by the Scale Cube.

Examples of the Scale Cube (Page 7)

• Using a banking application:

  • X-axis scaling: Duplication of the entire application across multiple servers for load balancing.

  • Z-axis scaling: Dedicated servers for different geographical customers with replicated applications.

  • Y-axis scaling: Decomposition of the application into microservices based on functionalities.

Microservices (Page 8)

• Advantages of microservices:

  • Allows individual scaling of each microservice using X- and Z-axis scaling.

  • Enables updates to the codebase of each microservice without affecting others, maintaining API compatibility.

Web Application Deployment (Page 9)

• Web applications may be hosted:

  • On-premises or via third-party data centers.

• Transitioned from physical servers to virtual machine clusters, then to container swarms for increased efficiency.

• Containers align well with microservices architecture.

Containers (Page 10)

• Containers virtualize the OS instead of hardware:

  • More efficient and portable compared to traditional virtual machines (VMs).

• Often deployed alongside VMs for optimal performance.

Containers vs. Virtual Machines (Page 11)

• Containers:

  • Package applications and dependencies, run as isolated processes.

  • Use less space (typically tens of MBs) and boot faster.

• Virtual Machines:

  • Simulate physical hardware with a hypervisor.

  • Require more resources (GBs) and take longer to boot.

Visual Representation (Page 12)

• Diagram comparison between containers and virtual machines illustrating architecture.

Containers and Microservices (Page 13)

• Microservices are often containerized for scalability.

• Kubernetes used for orchestrating, managing, and monitoring containers dynamically based on demand.

Load Balancing (Page 14)

• Distributes request loads among multiple servers to avoid overflow.

• Involves algorithms and dedicated hardware for optimal distribution.

Anycast (Page 15)

• A load balancing method sharing a single IP address among multiple hosts:

  • Unicast: One-to-one communication.

  • Multicast: One-to-many communication.

  • Broadcast: One-to-all communication.

Anycast Explained (Page 16)

• Anycast routes a request to the nearest (fewest hops) host sharing the IP address, facilitating efficient request handling.

Appropriate Use of Anycast (Page 18)

• Ideal for stateless connections:

  • Suitable for DNS, streaming services, and static content requests (e.g., images, stylesheets).

Content Delivery Networks (CDN) (Page 19)

• CDNs cache static content in geographically distributed locations to improve load times for end-users.

• Major CDN providers include Cloudflare, Google Cloud, and Microsoft Azure.

• Use of load balancing and anycasting techniques in optimizations.

Advantages of a CDN (Page 20)

• Latency Reduction: Serves requests from geographically closest data centers.

• Reliability: Ensures failover capabilities and redundancy across data centers.

• Security: Offers protection against DDoS attacks by distributing request loads.

Disadvantages of a CDN (Page 21)

• Potential privacy concerns for end-users, facilitating tracking and analytics by corporate entities.