Skip to content

Latest commit

 

History

History
43 lines (41 loc) · 3.18 KB

Cloud Computing.md

File metadata and controls

43 lines (41 loc) · 3.18 KB
Raw

Definition

Data centre hardware and software that vendors use to offer computing resources & services.

  • Pool of easily usable, virtualised resources (e.g VMs), and services (e.g [[Bigtable]], [[Dynamo]])
  • Typically easily scalable (dynamically provisioned compute resources)
  • Typically charged for usage (resources acquired and released on demand)

Some Providers are [[Google Cloud Platform]], [[Amazon Web Services]] and [[Microsoft Azure]].

Cloud Type Description
Public All hardware and software, and supporting infrastructure is owned by the provider (e.g [[Google Cloud Platform]], [[Amazon Web Services]] and [[Microsoft Azure]].) Resources offered over the internet.
Private Infrastructure is hosted by the company using it, all on-site on a private network.
Hybrid Data and applications shared between a private and public cloud setup.

Benefits

  • Speed - can provision new resources & services immediately, rather than waiting to order and set up own hardware
  • Elasticity - Only paying for usage, if demand is varied then compute can be scaled automatically, and the company only pays for usage.
  • Global Scale - Can be available worldwide, easily provision compute in different regions
  • Security - The provider is responsible for much of the security, and provides their own services to help with this.
  • Economies of Scale - Large datacentres have a lower cost per unit, providers can justify cost of micro-optimising their own hardware, software with huge scale.

Costs

  • Dependency of connection to the cloud provider (rather than using machines onsite for a purpose onsite), though this does not apply to consumer facing applications.
  • Vendor lock-in - if too dependent on a single vendor, a company needs to price-in the risk of increased fees.

Cloud Service Models

$$\underbrace{IaaS \to PaaS \to Faas \to SaaS}_{\text{less } \to \text{ more control}}$$

Infrastructure-as-a-Service (IaaS)

$$\underbrace{\text{App Data Runtime Middleware OS}}{\text{Customer}} \ \underbrace{\text{Virtualization Servers Storage Networking}}{\text{Provider}}$$ Provider leases servers & VMs, as well as storage, network, security infrastructure.

  • Customers write their own software, manage deployment to their leased machines
  • Provider pools resources to share serve multiple customers

Platform-as-a-Service

$$\underbrace{\text{App Data}}{\text{Customer}} \ \underbrace{\text{Runtime Middleware OS Virtualization Servers Storage Networking}}{\text{Provider}}$$ On-demand environment for development, testing and management provided.

  • servers, storage, network, OS etc
  • The current most popular model

Function-as-a-Service

Customer builds serverless functionality (e.g AWS Lambdas), deploys. Provider manages servers and infrastructure.

  • Customer only concerned with code, no exposure to what VMs, network etc is set up.

Software-as-a-Service

$$\underbrace{\text{App Data Runtime Middleware OS Virtualization Servers Storage Networking}}_{\text{Provider}}$$ Provider provides applications over the internet on demand.

  • All underlying app dev/infrastructure/setup is managed by the vendor
  • All maintenance completed by vendor. For example a company may use slack, office365.