Microsoft directly hosts a handful of big conferences every year, with the developer-oriented Build conference and infrastructure-oriented Ignite being to two that I pay the most attention to. The Microsoft Build 2019 conference just wrapped up and we have a couple important new announcements for the those of us working with the Microsoft Data Platform.
First and foremost, Microsoft publishes all of their sessions nowadays at no charge. So if you’re like me and you couldn’t attend, check out all the content at https://azure.microsoft.com/en-us/resources/videos/index/?series=build-2019. If you only care about the infrastructure announcements, head to the complete list of searchable announcements at https://azure.microsoft.com/en-us/updates/?updatetype=microsoft-build&Page=1.
On the Edge of Our Seats
As a dyed in the wool SQL Server guy, that’s always my first interest and we have plenty of fresh talking points here, the most salient being Azure SQL Database Edge. This new SKU for SQL Server answers new problems opened up by IoT computing enabling you to do lots of compute on the edge. Azure SQL DB Edge is a containerized package intended for both ARM and x86 powered devices, especially tailored for streaming and time-series data. It can be easily integrated into third-party IoT and edge products as well as Microsoft’s own edge models. It is now available in the Azure Market Place via private preview, so you will need to answer a brief questionnaire to get access.
I’ll talk more about Azure SQL Database Edge in next month’s column. But if you want to get a headstart, check out https://azure.microsoft.com/en-us/services/sql-database-edge/.
Serverless SQL Server
Behold the power of AWS into forcing Microsoft to choose a dumb name like Azure SQL Database Serverless. Clearly, I’m not a fan of the name, but the product itself is very coll. It’s a way to offer the power of SQL Server by billing per-second, based on memory consumed plus the number of virtual CPU cores used. Any time you pause the database instance, you are charged for the storage only. You can define idle times to automatically pause the database, while the service itself resumes as soon as there is a login activity detected. Details are at https://azure.microsoft.com/en-us/updates/azure-sql-database-serverless-compute-tier-is-now-in-preview/.
Lots of Machine Learning
There’s lots of progress on the Machine Learning (ML) front. For example, Azure ML Studio now integrates directly with Azure ML services, such that enabling you to launch visual designers within the workspace to take advantage of datasets and to leverage compute targets to train models. You can even join Notebooks to the Python SDK using some nice no-coding options.
Just as many other segments of the IT world are connecting themselves to the momentum of the DevOps trend, Microsoft is combining DevOps and ML into MLOps. MLOps is based on the GitOps workflow and helps data scientists and app developers to accelerate ML modeling to production. Using MLOps, you can track, version, audit, certify, and re-use assets in the ML lifecycle, as well as orchestrate and streamline the lifecycle. MLOps combines many Azure services such as storage, ML services, Kubernetes, DevOps, and more.
Finally, for the developers reading this column, ML.NET is now production ready. It is a free, cross-platform and open-source ML framework bringing ML into .NET applications. Using ML.NET, you can build apps to train, build and ship custom machine learning models using C# or F#. Read more at https://dotnet.microsoft.com/apps/machinelearning-ai/ml-dotnet.
Hyperscale That Mountain
Back in the January 2019 edition of this column, I told you about the Azure SQL Database Hyperscale public preview. It is now generally available for SQL Server and is in public preview for Azure PostgreSQL Database. Read more details at https://azure.microsoft.com/en-us/blog/get-high-performance-scaling-for-your-azure-database-workloads-with-hyperscale/. Hyperscale is tuned especially for online transaction processing (OLTP) workloads and has very impressive performance capabilities, for example, it can restore mult-terabyte databases in mere minutes.
Azure SQL Data Warehouse
Azure SQL Data Warehouse keeps getting better. It now has several new features, such as Result-set Caching, Materialized Views, and support for sorted columnstore indexes. Result-set Caching facilitates nearly instantaneous query processing and greater concurrency. Similarly, Materialized Views helps speed complex business logic for reports and analytics by storing the view result sets in the database and keeping the results up-to-date as data changes. Read more at https://azure.microsoft.com/en-us/blog/analytics-in-azure-remains-unmatched-with-new-innovations/.
On the ETL side, the Mapping Data Flows capability in Azure Data Factory is now GA by providing a GUI on top of Azure Databricks (Microsoft's managed Spark service). Mapping Data Flows is fast, due to the way it performs transformations in memory without having to land data and then use separate compute engine to process.
Bits and Bobs
Microsoft also announced some nice connectivity and migration tools, especially around my favorite open-source database, PostgreSQL. For example, there is a new extension for PostgreSQL called pg_auto_failover which enables you to automate server failovers for HA and DR purposes. Since pg_auto_failover is released under the PostgreSQL license, I’m sure we’ll soon see community members adding cool community extensions. In addition, Microsoft has released a new Oracle-to-PostgreSQL migration assistant.
I look forward to hearing more about what’s on your mind. Drop me a note in the Comments, send me an email at kevin_e_kline@yahoo.com, or reach out on social media to me at @kkline (Twitter, LinkedIn, and Facebook).