A company has moved all on-premises workloads into a public cloud. After some time, the cloud engineer starts noticing time drifts on the VMs and suspects an NTP issue. Time drifts were not an issue…

In cloud environments, it is best practice to configure all virtual machines to use Coordinated Universal Time (UTC) and let the hypervisor or cloud provider's host manage the time synchronization to a reliable NTP source. Each VM can then display the local timezone based on UTC, preventing time drifts that often occur when VMs attempt to manage their own time or use internal NTP servers not optimized for virtualized environments.

Pointing cloud workloads back to an on-premises NTP server introduces latency and potential network connectivity issues, making time synchronization less reliable than using cloud-native NTP services or host synchronization.

While pointing to a stratum 2 NTP server within the public cloud provider is a better option than on-premises, the fundamental issue in virtualized environments is often related to the interaction between guest OS time and hypervisor time, making the consistent use of UTC and host synchronization the more robust solution.

Configuring all VMs as NTP servers is highly inefficient and creates a complex, fragile NTP hierarchy within the cloud, which is prone to further time synchronization issues and unnecessary resource consumption.