Azure Updates: South African government contract; TensorFlow; Azure billing; Stream Analytics; S/4 HANA disaster recovery

As October drew to a close and November got underway, Microsoft issued relatively few Azure updates while the partner channel remained rich with new announcements. Writing on Investopedia, finance and markets reporter Shoshanna Delventhal argued that even after winning the JEDI contract, Microsoft may still need to catch up with Amazon. For now, AWS continues to lead the market with a 47 percent share, but Azure appears to be catching up with 59 percent growth rate last quarter compared to Amazon's 35 percent. While JEDI garnered plenty of market attention, Microsoft also enjoyed a major win in South Africa, signing a memorandum of understanding with the State Information Technology Agency to rehost South African government infrastructure to Azure.

With the launch of TensorFlow 2.0, Microsoft program manager Gopal Vashishtha explained how Azure Machine Learning can be used with TensorFlow workloads. In particular, the Machine Learning SDK for Python offers a TensorFlow estimator, which helps with running TensorFlow training scripts on a wide variety of target machines.

Microsoft is changing the unique identifier for Azure billing accounts, although the shift only affects accounts under the Microsoft Customer Agreement and not the Enterprise Agreement. When using the newest version of APIs such as Invoices, Payment methods, Transfers, Cost Management or Budgets, users will see new values.

Among geographic updates Site Recovery is available in the Norway East and Norway West regions. The Azure team is also previewing Blob Storage on IoT Edge integrated with Event Grid and notified users that Azure Sphere will be generally available by February 2020. When Sphere comes online, it will offer new ways for customers to rapidly secure IoT devices.

Throughout the past year, machine learning has remained a key focus for the Azure team. On November 4, new features are rolling out for Stream Analytics, including online scaling, which means users no longer have to stop a job if they need to change SU allocation. The rollout also includes Managed Identity authentication on Power BI, extensibility with C# and C# custom de-serializers, Visual Studio Code debugging and availability on Azure Stack. The move comes as Microsoft works to tout the capabilities of Azure Machine Learning, in a new video with ML team member, Chris Lauren, discussing CI/CD model retraining, packaging models with MLOps, and SDK and container service integration details.