Reliability and Predictability

Source: https://www.youtube.com/watch?v=kD2YqdDaO1w&list=PLlVtbbG169nED0_vMEniWBQjSoxTsBYS3&index=7

Nodes and racks can fail, but in the cloud it an heal itself

Reliability

• For example, if your VM goes down, Azure can cover it and redeploy that VM elsewhere

• With storage accounts, there are at minimum 3 copies of your data

  • With databases, you might have additional replication with log replay on a separate server

• Auto Scale is the ability to add additional resources to your app and deallocate those resources after the fact

• SLA

  • Each service has a different SLA

• Design for failure is something that needs to be considered when developing on Azure

  • Your server might be active->active, active->passive, etc, but you need to design for that

• Azure monitor allows you to create alerts and action groups so that you can keep an eye on those resources

Predictability

• Many different SKUs are available in Azure

  • Certain memory, CPU, throughput, etc

• Behavior

• Templates allow you to quickly deploy new assets using a standardized format

  • This also allows for automation to perform maintenance activities

  • Devops