Throughout the meeting, I was impressed by the coordination and planning as hundreds of vendors hosted booths and demonstrations and lectures. Microsoft of course had multiple booths, featuring new tech like Azure Sphere for IoT, Azure Data Box for huge file transfers, and VR demos built on Azure services. Channel 9, Microsoft’s video site for developers, recorded segments with authors, presenters and commenters throughout the week. The giant video wall lounge, three stories tall with 12 channels of breakout sessions playing, was constantly full with attendees relaxing in theater seating or playing Xbox games.
Here are some of my takeaways from this week-long event.
Machine Learning DeliversAI and machine learning were front and center at the conference, starting with the keynote address by Microsoft CEO Satya Nadella. Cognitive Services have made giant leaps forward in the last year.
- Object classification, facial recognition, and OCR are simple REST API calls for images.
- Translation, key phrase extraction, and natural language Bot frameworks create seamless interfaces for communication.
- QnA Maker combines a list of facts and questions with translation and natural language processing to create a Q&A service.
- Video Indexer combines the vision services and language services to automatically tag, translate, transcribe, and bookmark videos.
- Analytics services like Azure Stream Analytics have integrated sentiment and anomaly detection algorithms based on Microsoft Machine Learning.
Big Data TechnologyBig Data, or Analytics at Cloud Scale, has really matured in the last year in Microsoft Azure.
- HDInsight continues to provide a powerful and scalable platform for Hadoop and Apache Spark and Storm data scientists. The cluster management works well, and Zeppelin notebooks provides easy investigations into huge datasets.
- Microsoft improved on the Big Data processing pipeline with the introduction of Azure Databricks. Databricks brings the ease of SaaS deployment to streaming analytics, with easy integration to Azure Storage and Databases.
- Cosmos DB has moved to the center of PaaS Big Data architecture. The new Change Feed API provides a path for either Lambda or Kappa processing architecture. Cosmos DB includes global distribution and speed for many types of data storage and queries. Microsoft added a lower minimum tier for easier entry to usage. Cosmos DB integrates well with cloud-based platforms using Azure Front Door, a globally distributed load-balancer, firewall, and router for Internet services.
SQL Server for Every NeedMicrosoft has not slackened their commitment to SQL Server, even as they introduce options like Cosmos DB, PostgreSQL, and MySQL.
- SQL Server 2019 can connect to a Hadoop File System (HDFS) store using Apache Spark and execute Jupyter notebooks.
- Azure SQL Database Hyperscale disconnects the SQL query engine from storage, which can be directly accessed from Blob Storage. This allows SQL DB to grow beyond the 4TB limit.
- Azure SQL Data Warehouse continues with separately scalable compute and storage nodes. New version 2 adds an inline caching service to increase performance.
- Azure SQL Database Managed Instance now has full compatibility with on-premises SQL Server 2016.
As you can tell, the conference showcased a number of new developments at Microsoft, and included many opportunities for talking with Microsoft team leads and industry partners. An oft-repeated call was for feedback on cloud services. Email and blog and support requests are followed by project managers and the teams are very responsive to the user base.
It was a valuable meeting, and the teams back here at Silverchair have enjoyed discussing how we can incorporate various learnings into our offering.