The Microsoft Build conference is developer heaven. It’s also a great place to learn what’s coming for the rest of us in the Microsoft world, and this year’s conference, virtual though it was, was no different. Microsoft introduced several new features for its cloud service, Azure, leaving no doubt where its priorities now lie.
Take a deep breath and let’s have a look.
First up, analytics
Microsoft describes Azure Synapse as “Azure SQL Data Warehouse evolved” – not very helpful to most of us until you pick up on the tidbit that it’s an analytics service that brings together enterprise data warehousing and big data analytics. At Build, the company made two announcements around Synapse: First, Azure Synapse Link for CosmosDB (also coming soon for Azure SQL, Azure Database for PostgreSQL and Azure Database for MySQL), which will allow analytics to be run on live operational data. Second, several new Synapse features were announced, including a unified experience with Synapse Studio, integrated SQL and Spark engines, serverless data lake exploration with Synapse SQL, a shared metadata store and built-in Power BI authoring.
On the infrastructure front, we got the public previews of Azure Arc-enabled Kubernetes for management of clusters from Azure across datacentres, multicloud, and Azure Stack Hub, and of support for SUSE Linux Enterprise Server as an Arc-enabled server. Azure Stack Hub itself received some love, with previews of Fleet Management (one view of all Stack Hub deployments), ManageIQ support to manage hybrid IT, Azure Kubernetes Resource Provider on Azure Stack, and GPU partitioning using AMD GPUs.
Most of the data-related announcements were wrapped around Microsoft CosmosDB, the company’s fully-managed NoSQL database service. New capabilities include point-in-time backup and restore, version 4 of its Java SDK, “bring your own key” for enhanced end-to-end encryption, autoscale provisioned throughput offering SLA-backed 99.999 per cent availability and single-digit millisecond latency. It introduced a new pricing model, CosmosDB Serverless, to handle workloads with intermittent traffic and bursty workloads.
Microsoft also launched new capabilities for Azure Database for Postgre SQL and Azure Database for MySQL and contributed more technology to the Postgre SQL open source project.
In addition, Azure SQL Edge for connected, disconnected or semi-connected environments entered public preview, offering streaming, storage, and AI in a small-footprint container suitable for edge devices.
Finally, Microsoft announced that the Azure NetApp Files service has achieved 99.99 per cent availability.
Artificial intelligence (AI) got a considerable amount of attention, starting with Responsible Machine Learning (ML) in Azure Machine Learning, which includes model interpretability and fairness assessment (looking for bias in the model), privacy protection, and new data control features.
Azure Cognitive Services adds Personalizer apprentice mode that learns alongside existing services when deployed, only interacting with users when the Personalizer service’s level of confidence matches that of the existing solution.
Microsoft announced that Azure’s speech services both have improved accuracy and work in additional languages, with Speech to Text expanding to 27 new locales, and Text to Speech extending to 11 new locales. Language Understanding and Text Analytics sentiment analysis are now available in containers so they can be deployed anywhere from cloud to edge.
QnA Maker, Azure’s automated API service that builds questions and answers from semi-structured content such as manuals, FAQs and documents, now lets users collaborate on the knowledge bases it creates and allows content managers to control the formatting of responses to users.
Winding up the AI bits, the company announced previews of two new features for Azure Cognitive Search, its AI enriched cloud search service: custom search ranking integration into Cognitive Search, for customers who have built their own search results ranking systems, and integration with the Azure Machine Learning model registry and machine learning capabilities to ensure the latest versions of models are automatically integrated.
After a slow start, Microsoft has dived into the Internet of Things (IoT) with enthusiasm. It’s got some catching up to do, but current announcements are a nice start. Azure Digital Twins, a platform for creating digital representations of physical devices, environments, or processes which was announced for public preview in 2018, will offer expanded capabilities this summer, including support for OPEN Modeling Language (based on JSON-LD) to improve ease-of-use for developers, a live execution environment, integration with Azure IoT Hub and other services, and query APIs.
And speaking of developers – the Azure IoT Developer Specialty certification has emerged from beta and is now available to all.
Azured out yet? There’s a ton more, but let’s look at just one more announcement: Project Bonsai.
Project Bonsai, now in public preview, is Microsoft’s machine teaching service to create and optimize intelligence for industrial control systems. It enables input from experts as well as knowledge from data, and lets users understand how the AI agents work and helps them debug when the agent doesn’t work as expected. Microsoft is also announcing Project Moab, a new open source balancing robot to help engineers and developers learn how to build real-world autonomous control systems with Project Bonsai. Customers can 3D print the robot, and availability for purchase will be announced later this year.