Long-running operations (e.g. Create lab, start VM, delete custom image) are being delayed by up to 2 hours. The impact appears to be limited to the following regions: Australia East, Brazil South, Canada East, East US, East US 2, Japan West, Korea South, North Central US, South India, Southeast Asia, UK South, West Europe, West US 2.
The incident seemed to start at around 10 AM Pacific (18:00 UTC), is currently ongoing, and the team is investigating and working to mitigate.