Azure and IP Persistence - VM,Web/Worker Roles

If you already worked with Windows Azure, especially with worker roles, web roles or VM I suppose that you noticed that Azure don’t guaranty a static IP of this machines. Because of this it is not recommended to access or connect different Azure resources based on the IP (internal IP).
But in the same time, there are moments when the IP is persisted. For example when you restart the resource. In this blog post I will try to explain when the IP of the resources will not change and will remain the same after different actions.

Case: Upgrade
Let’s suppose that our resource is restarted after an update. In this case the original IP of it will be persisted.
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11 (restarted)
After:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11

Case: Hardware fail
When we have a resource that I take down because of different situations (for example hardware fail) the original IP of the machine will be persisted after the instance is re-deployed. Even if the instance is moved on a different physical device.
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11 (hardware fail)
After:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11

Case: Resource reboot
In the moment when a resource is rebooted the original IP will be persisted.
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11 (reboot)
After:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11

Case: Scale up with a new image
When we need to scale up, allocating a new image of our resource a new IP will be allocated to this resource. The existing images will remain with the same IP as before.
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
After:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
VM_IN_2 – 10.0.0.11

Case: Scale up after a scale down
After a scale up and a scale down, the next operation of scale up will not create a new image of the resource with the same IP as last time when we scale up.
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
Scale up:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
VM_IN_1 – 10.0.0.12
Scale down:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
After (scale up second time):
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
VM_IN_2 – 10.0.0.13

Case: Resource is de-allocated (stop de-allocate)
When we have a resource that is de-allocated we cannot assume that the current IPs will be persisted until the resource will be allocated again. Allocating the resource again will use different IPs
Example:
Before:
VM_IN_0 – 10.0.0.10
VM_IN_1 – 10.0.0.11
De-allocate
After (allocate resource again):
VM_IN_0 – 10.0.0.12
VM_IN_1 – 10.0.0.13

If we have situations when the IP of the resource is the same, there is another story with MAC address. Don’t assume that if the IP will be the same, than the MAC address will persist also. For example for the hardware fail case, the MAC address will change because another VM will be used.