What are the effects?

Here, I first have to take a small step backward. The majority of planned maintenance will not affect the operations of the platform (for example, storage infrastructure updates or physical network infrastructure updates).

With other updates that affect your VMs directly, things look different. This includes the following:

  • Host security and compliance patching
  • Updates for the Azure host agents
  • Updates for the network agents
  • HW decommissioning
  • HW maintenance
  • Host OS rollout

Since each of these updates places specific requirements on the platform, two subtypes of planned maintenance have evolved:

  • VM preserving maintenance (also known as preserving host updates): The VM is placed in a Paused state, and will be resumed within 30 seconds. Used memory, network connection, and open files are kept in their respective state. The maintenance type, VM preserving maintenance, is used for updates in the areas of host security and compliance patching, Azure host agents and network agents.
  • VM restarting maintenance: Basically, this type of maintenance has one thing in common. Each affected VM will be rebooted in some way. Nevertheless, there are two subtypes here:
    • VM reboot: The VM will be restarted. The state of memory, network connections, and open files are lost.
    • VM redeploy: The VM is moved to another host. The state of memory, network connections, open files and the transient drive info are lost.

The maintenance type, VM restarting maintenance, is used for updates in the areas of HW decommissioning, HW maintenance, and host OS rollouts.

Best practice: Although Microsoft is currently trying to simplify the planned maintenance process for the customer (for example, a VM restart event is not scheduled to happen more than once a year), you should secure your applications by letting it run on at least two instances at the same time.

As I've already written, planned maintenance is the only predictable downtime. For many, the question arises of how do I do that?

Here is some brief guidance:

  1. Open your Azure management portal at https://portal.azure.com.
  2. In the navigation area of the portal, click on Help + support.
  3. This opens the Help + support dashboard, as shown in the following screenshot. In the navigation area of the dashboard, click on Planned maintenance.
  4. If a scheduled maintenance is pending, you can see it in the now open list (No maintenance events are scheduled):

It is important to know that you do not need to visit the Help + support dashboard regularly to avoid planned maintenance. Thirty days before a planned maintenance, you will also receive a notification about the platform's notification service, as shown in the following screenshot:

Here is an anecdote to end with and maybe a reason to think carefully about the issue. In February 2012, it was part of a planned maintenance to import a faulty update (the update had the leap year excluded). The result? The operation of most of the Azure data center was canceled for a long time.

However, I do not want to end my remarks without at least once going into the second approach of availability definition. There is an interesting solution for this, which I will introduce to you now.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
18.218.78.102