Terraform Backends

• Terraform by default when applied to create infrastructure will create a local state file, so when multiple users execute the terraform on multiple machines it leads to creation of multiple target infrastructures which is not idle.
  
• The solution for this is if we can store the state file in some common location which is referred as Backend, then even if multiple users try to apply it leads to same infrastructure.
  
• To avoid multiple users writing to the state file at the same time, terraform implements locking
• Refer Here for the official docs of the backend
• Terraform supports the following backends
  
• Lets first implement the backend using azurerm and then we will implment s3 based backend

Azurerm Backend

• Lets create a resource group
  
• Now create a storage account
  
• Once the storage account is created, create a container
  
  
• Now lets add the backend to our terraform
  
• Now execute terraform init to initialize the backend
  
• Now lets execute terraform apply -var-file="experiment.tfvars" -auto-approve
  
• To simulate the multi user scenario, let create a new linux vm and install terrform over there and see parallel executions
• Now lets apply terraform on linux system
  
• Before the state has been applied lets try to simulate apply by other user i.e on my windows system
  

AWS S3 Backend

• Refer Here for official docs
• Create an s3 bucket
  
• Lets create a dynamodb table for locking purposes
  
  
  
  
• Now initialize
  
• And Apply
  
• As this user has acquired the lock which will be released upon completion of terraform execution till that moment no other user will be able to create infra.