Azure Service Endpoint Demo

Azure Servie Endpoints allow you to extend the identity of your virtual network subnet to certain Azure services such as SQL or Storage. Extending identity of the network allows the Azure service to only accept traffic from clients deployed within that subnet. This allows the service to deny requests from other services and apps using the Azure backbone, reducing the surface area for an attack.

This repository contains a terraform script to setup a virtual network, with an Ubuntu VM deployed to the 'front' subnet that writes blob to an Azure storage account. The 'front' subnet has the Microsoft.Storage service endpoint enabled. The storage account has a network rule applied to allow requests originating from the front subnet to hit the storage endpoint.

Application

To simulate an real application living in our front subnet, an Ubuntu VM named app-endpoints-demo-wu2 uses the Azure CLI to write blobs to our storage account. This virutal machine is setup as part of the build\network.tf script. A public ip is associated to the NIC to allow us to ssh into the machine with a username and password.

Building the Network

Before setting up the environment through the network.tf script, set the locals variable for public_ip to your public IP. This ensures you can access the storage account from your developer maching.

locals {
  public_ip = "XX.XXX.XXX.XXX"
}

To setup the environment:

Update Terraform AzureRM Provider Settings

At the top of the network.tf script. You will need an App registration in Azure Active Directory with a client id and secret that has contributor access to the subscription. See the Service Principal Client Secret documentation.
```
provider "azurerm" {
    subscription_id = "SUBSCRIPTION_ID"
    client_id       = "CLIENT_ID"
    client_secret   = "CLIENT_SECRET"
    tenant_id       = "DIRECTORY_ID" # DirectoryID
}
```
Plan and Apply the Terraform script

Sets up the network, storage account, subnets, virtual machine and additional infrastrucre to test the network.

Plan
```
terraform plan -out out.plan
```
Apply
```
terraform apply out.plan
```
SSH into the VM

You first need to get the public ip of the virtual machine on the network.
```
az network public-ip show -g rg-endpoints-demo-wu2 -n app-publicip-endpoints-demo-wu2 --query ipAddress
```
Using the public ip, SSH into the virtual machine. The demo uses a username and password to SSH. You wouldn't want to use this setup in a production environment. The password setup in the network terraform script is T3x4stoast.
```
ssh azureuser@PUBLIC_IP_ADDRESS
ssh azureuser@52.191.167.195 (ex)
```
If you see an erro fro the az cli, you may need to login with az login first for your test account.
```
{
    "error":"invalid_grant",
    "error_description":"AADSTS40016:" ...
}
```
Install the Azure CLI

The build\azurecli.sh script installs Azure CLI into your virtual machine. You can execute that script remotely from my repo. See executing remote scripts on StackOverflow
```
curl -s https://raw.githubusercontent.com/MikeLarned/azure_vnet_service_endpoint_demo/master/build/azurecli.sh | bash
```
Storage Account Key

To write blobs to the storage, we need the storage account key. You can access this key through the Azure CLI. Key1 or Key2 should work.
```
az storage account keys list -g rg-endpoints-demo-wu2 -n saendpointsdemowu2
```

Enable Request Logging in Azure Storage

You want to enable minute request logging for saendpointsdemowu2 storage account. This writes logs to a folder called $logs. These log files allow us to view inbound ips for storage account requests.

az storage logging update --log "rwd" --retention "7" --services "b" --account-key "KEY" --account-name "saendpointsdemowu2"
az storage metrics update --log "rwd" --retention "7" --services "b" --api "true" --minute "true" --hour "true" --account-key "KEY" --account-name "saendpointsdemowu2"

Write Blobs to Storage

From the VM you SSH'd into, execute the storage.sh script remotely. This script generates a blob with a GUID name and writes it to the storage account. You need to ensure the last parameter to the script is the key you received in Step 5 of the setup.

    curl -s https://raw.githubusercontent.com/MikeLarned/azure_vnet_service_endpoint_demo/master/build/storage.sh |  bash -s "saendpointsdemowu2" "KEY"

Reviewing the Logs

Log files are located in the $logs folder inside of the storage container. Open the folders until you get down to the days and hours logs. For example, my log file was located in $logs\blob\2019\03\12\0200\000000.log. Open the log file and look for a row containing 1.0;2019-03-12T02:58:14.2840912Z;PutBlob;Success;. If you split that row by the ; delimiter, line 15 or 16 should contain the inbound IP of the request.

0
10.200.10.5:53262
2018-03-28

Here we can see 10.200.10.5, the private IP assigned by my virtual network back subnet.