Windows Azure IaaS Host operating system update disclosure

In customer service cases created by customers, http://www.aliyun.com/zixun/aggregation/13357.html ">azure VM accidental reboot is a common problem, and customers ask customer service to determine the cause of the reboot." Hopefully, the following detailed instructions will help you understand the reason for the Azure VM reboot.

Windows Azure updates the host environment every six to 2-3 months to ensure that it always provides a secure environment for all applications and virtual machines running on the platform. This update process may cause your VM to reboot, resulting in an outage of applications/services hosted using the virtual machine feature. There are no options or configurations to avoid these host updates. In addition to platform updates, when a host server failure is detected, Windows Azure service is automatically repaired, and VMS running on that server are migrated to other hosts. When this happens, you lose the connection to the VM during service repair. After the service repair process is complete, when you connect to the VM, you may find an event log entry indicating that the VM was restarted (normal or abnormal). Therefore, you need to configure your VM to handle these situations, thereby avoiding application/service downtime.

To ensure the high availability of applications/services hosted in Windows Azure virtual machines, we recommend using multiple VMS in the availability set. VMs in the same availability set are placed in different fault domains and update domains, ensuring that scheduled updates or unexpected failures do not affect all VMS in the availability set. For example, if you have two VMS and you configure them as part of the availability set, when a host is updated, only one VM stops at a time. This provides high availability because you have a VM that can process user requests during the host update process. Mark Russinovich published a very good blog article detailing information about the Windows Azure host update. For more information about managing high availability, click here.

While the availability set provides high availability for VMS, we are also aware that proactively notifying customers about planned maintenance is a common feature that needs to be used, especially if your workload is running on a single VM and not configured for high availability. Although this type of proactive scheduled maintenance notification is not currently available, we encourage you to provide comments about the topic, and then we can feed this information back to the product team.

[UPDATE] Currently, if there are planned maintenance, we will send a notification to the VM of a single instance. However, only the account administrator may see this notification.

Windows Azure IaaS host operating system update disclosure

Special thanks to Sri Harsha for taking the time to review this blog!!

In this blog, we will discuss the following issues:

Why does Windows Azure update the host operating system?
How do I update the host operating system?
What is a set of availability?
Why create an availability set to make your application highly available?
Some resources to help you create and manage high-availability applications in Windows Azure virtual machines
More information
Why does Windows Azure update the host operating system?

Windows Azure deploys updates to the host operating system at approximately the monthly frequency. This ensures that Windows Azure provides a reliable, efficient, and secure platform to host your application.

How do I update the host operating system?

The host operating system update on the Windows Azure platform differs from the way you update your PC or server running on the Windows operating system. For Windows Azure, when an image that contains all the latest updates and hotfixes is deployed to all servers, Fabric Controller instructs them to reboot and start the newly deployed image. As a result, Windows Azure host operating system updates only take time to start from a new image, rather than Windows updates. Typically, it takes only 15-20 minutes to complete this host operating system update process.

What is a set of availability?

When you have two or more VMS performing the same task (for example, two or more WEB servers), you can create a set of availability that contains the two VMs. Creating this availability set makes your application highly available and also qualifies you for an SLA of 99.9% uptime.

Why create an availability set to make your application highly available?

When you create an availability set, you can instruct the Fabric Controller, all VMs in the availability set perform the same function, and do not have to stop at the same time to perform scheduled maintenance.

In this case, the Fabric Controller will intelligently place these VMS in different update domains (UD). These UD are logical classifications that help FC ensure that all VMs in the same as are not shut down at the same time during any scheduled maintenance. This ensures that VMs are always available to process requests.

Note：

Test/monitor to ensure that even if the number of VMs running the workload is reduced, sufficient performance can be provided so that if one or more VMs are not available, your service will not be adversely affected during the scheduled maintenance period.
If you use endpoints to allow incoming traffic, make sure it is in a load-balanced state. (see "Creating High Availability workloads with Windows Azure" below.) ）