AWS Classroom Series – 05/Oct/2021

Need for Compute

• For any organizations, applications mean business and to run the applications we would need servers, operating system etc
• Consider the below architecture of the application
• Generally in the traditional enterprise to realize this application we have 2 ways
  • Physical Infra
    • As shown in the above image, we would have 3 physical servers and we install necessary os & install softwares and deploy applications
  • Virutal infra:
    • Organizations use hypervisors (like vmware ) to create multiple virtual machines on the same physical server to effectively use hardware resources
• Constraints of On-Premises Infrastructure:
  • Consider this application is used by customers in UK, we need to physical infra in UK. Business over different parts of the globe might be a costlier choice
  • Scaling servers is possible but always dependent of free resources
  • Datacenters can go wrong, Disaster Recovery needs to be setup.
• Cloud: Cloud Service Providers have infrastructure across the globe and we can use that infra to realize our applications.
  • With cloud global applications are easier to create
  • Scaling up or down (Elasticity) is easier to setup
  • Business continuity and Disaster Recovery is also easier than physical infra
  • In Cloud we are responsible for virtual infra not physical.
  • When we create a virtual machine, we assign
    • CPU and RAM from Physical server
    • virtual disk from some physical storage
    • some operating system to the virtual machine
    • some virtual network to get ip address to virtual machine
  • If we get one level deeper. When VM is running
    • it uses CPU, RAM
    • it uses virtual disk
    • it uses virtual network
  • When VM is stopped
    • it needs virtual disk
  • For the above mentioned reasons, when virtual machines are running you will be charged hourly for cpu/RAM + virtual disk cost + os license cost (hourly) and if you stop the vm you will still be charged hourly for virtual disk.
  • Even in the case of AWS where we call Virtual machines as EC2 instances the pricing is applied as mentioned above.