PASS Summit 2018 - Day 1 Keynote Announcements

Good morning from Seattle, at the Summit for the Professional Association of SQL Server.

I’m lucky enough to be sitting at the blogger table this morning, watching the keynote of announcements.

I’m following along with the keynote and making notes on slides – and I’ll share those slides with you below.

Don’t feel like reading or rewatching the presentation?

Join me on Tuesday, Nov 20th, with Grant Fritchey and Steve Jones in a webcast to sum up everything we learned at PASS Summit!


Keynote summary in slides

SQL PASS Summit 2018 from KendraLittle2

Text from the slides

PASSION award winner: Michael Johnson

Rohan Kumar

  • Hybrid cloud is the true enabler for digital transformation

  • AI is helping MS customers

    • Understand their customers and better meet their needs
    • Improve their operations
  • Critical to build training model on data that spans the hybrid estate

  • CTP 2.1 has been released

    • Monthly releasesUsing Azure to get feedback on a constant basisRequest for engagement from you

Conor Cunningham and Bob Ward

  • Demo: Removing page latch waits in tempdb
  • Behind the scenes, it’s using Hekaton in system tables in tempdb to speed this up
  • Not yet in preview
  • Will be in SQL Server 2019

Asad and Nellie

  • Demonstrating Azure Data Studio and Data Clusters

  • Use Python and Notebooks

  • Can query HDFS using native features in SQL Server engine to read that data

  • Access multiple data sources through SQL Server

More from Rohan Kumar

  • Azure Database Migration Service
  • Near zero downtime
  • Migrate at scale
  • Optimize IT Infrastructure
  • Azure SQL Database – 5 million active at any give time, a pedabyte of telemetery data every day
  • Microsoft is using machine learning against the data
  • Managed Instances
  • Big push to get customers here
  • General Availability of Business Critical SKU starting December 1
  • Azure SQL Database Hyperscale
  • 4 TB limit on initial implementation
  • Have been working on rearchitecting
  • Scales out storage over various nodes
  • Has abilities for fast point in time restore
  • Scale storage and compute independently
  • Accelerated DB Recovery & Machine Learning Services
  • Azure SQL Database
  • Goal is to make sure that no matter what happens, recovery happens in “constant time”
  • Machine Learning workloads in Azure SQL Database – enables migration for these features

Lindsey Allen

  • We don’t have fancy machines
  • We have 2 socket machines
  • Most customers don’t like figuring out sharding with partitioning keys – whatever key you choose is wrong
  • This is going to happen automatically behind the scenes
  • Now they take snapshots and restore from those – not a size of data operation
  • Accelerated Database Recovery
  • How long will it take to restore?
  • Painful for customers
  • Painful in Azure as well
  • Can aggressively truncate the log – even in full recovery mode

Deborah Chen

  • Multi-Master Replication in CosmosDB
  • Azure Cosmos DB
  • Apps can read from any region and write to them as well
  • Drawing app
  • Data being replicated to Japan in almost real time
  • Go to instance in Japan and drawn, and it app

John Macintyre

  • SQL Data Warehouse competing
  • 30% less expensive than Amazon Redshift
  • Fastest cloud DW based on benchmarks
  • Demo with 6 trillion rows of data
  • TPC-H Benchmark running
  • Processing over a trillion rows a second
  • Demo: prioritizing workloads in SQL Data Warehouse
  • Demonstration of queries being queued, waiting for system resources on a busy system
  • Connect to a dashboard using a special service account
  • That work gets prioritized and jumps the queue
  • Feature name: workload management

Ariel Pisetzky of Taboola

  • Azure Data Explorer
  • Used across Microsoft internally for several years to explore logs and do analysis
  • Ingest unstructured and semi-structured data
  • Not a lot of prep work to do quick queries
  • Demo: exploring data
  • Can query and show results in graphs or tables
  • Using the tool to report on the customer experience
  • “What is the 5% of the slowest recommendations we provide?”
  • • Identify spikes where they are above 800 ms
  • • Using queries to zoom in and identify who is impacted

Patrick LeBlanc

  • Dataflows
  • Point Power BI to any data lake store
  • It will figure out connectivity and transformations for the scenario
  • Big customer request
  • “Self-service for big data” stored in the data lake
  • SSRS Reports in Power BI
  • Reporting Services – things went quiet for a long time
  • Paginated reports in Power BI Service

Deepsha Menghani

  • Demo: Shell health & safety portal
  • Predictive alert
  • Azure database notebook
  • Need to have data to feed and train – and you can’t just burn down a gas station – she checked
  • Can do a keyword centric search for images to feed into the model
  • Image classification – databricks lets you build on existing models
  • Live demo: image detection
  • Hold a cigarette in front of a camera
  • Image identified and detected
  • Notification suggesting stopping a pump

And we’re done! Goodbye from Rohan