The Latest Azure Updates: VMs; Machine Learning; IoT data management; Azure HDInsight; Azure Stack

Conferences and big announcements packed the calendar for much of May and into June. Microsoft took a few days as the middle of the month approached to build off of some of its previous announcements, offering users more details, even as HL7 FHIR DevDays attendees gathered on the Redmond campus to discuss interoperable healthcare data.

Maxim Lukiyanov, principal program manager for Azure Machine Learning looked into the Notebook Virtual Machine, announced in May. He stated that the VM is a workstation for Python developers that reconciles a conflict between data scientists who want direct access to lower level CUDA drivers and admins worried about weakening security controls. Notebook access is now secured "out of the box" with Azure Active Directory and HTTPS. It is also highly customizable—Lukiyanov gave the example of data scientists setting up an NVidia V100 VM to debugging a new neural network architecture. Senior program manager Roope Astala built on the data science theme, describing the open source MLflow project. Developers are able to take existing code, instrumented as MLflow and put it into Azure Machine Learning for a training run.

Among geographic updates, Microsoft extended Azure Monitor for VMs support to West US 2, even as the VMBoundPort data set, with Active Ports and Open Ports workbooks went out to all regions with Log Analytics workspaces.