I’ll be updating this for the next hour or so…and done! This went FAST, and it was content intense. They went fast, and they needed a bit more storytelling….for people who aren’t native BI guys, for example, an awful lot of the demos and significance was lost, and the applause noticeably suffered.
Having said that, there were lots of demos- which is GREAT – and lots of technology, and very VERY little third party commercials. Great marks, there.
Welcome to my blog, where we’re (I’m) writing a blog live from the PASS Summit day 1 keynote blogger’s table! That means I have a name card and a table, which is nice. And that you get info from someone on the front lines (technically we’re in the center of the room, but still). Let’s see what’s up this year.
Up first: Adam Jorgenson! This is PASS Summit’s 19th year. I’ve been coming nine years or so, so the more I attend, the more of an old timer I get to be!
This is being live streamed on PASS TV, and so will more content from the conference, which is (seriously) fantastic.
By the way, half of the fun of the keynote is watching the #PASSSummit hashtag on Twitter…just saying…
Adam’s talking up PASS, which really does facilitate a lot of really great networking and education. He’s also thanking past and present volunteers and board members.
Next up, keynote speaker Rohan Kumar – general manager for DB systems engineering at Microsoft.
Keynote speaker: Rohan Kumar
But first, a video!
Okay, Rohan’s up. “Data cloud and AI are three of the most disruptive technologies of our time….”
He mentions the quick release schedule of SQL Server 2016 and 2017. They’re working on keeping up with change fast. He’s going to jump into new features in a minute! This is what most of the nerds at the blogger table are waiting for…
The modern data estate is a hybrid model between on-prem and cloud databases. It’s really, REALLY huge that they recognized this publicly! Most of what we hear out of Microsoft is all cloud, all the time. He asked the audience to raise their hands if they felt they could move completely to cloud computing, and retire their data centers. Two people raised their hands. We have hundreds if not thousands of people in here.
Slide: “SQL Server 2017 on Linux, Docker, and Windows server.”
Conor Cunningham and Bob Ward are invited to the stage, and there are actual sighs of delight from the blogger table. We love an SSMS demo.
They’re going to show off “5x performance with no code changes”…using scalable persistent memory, “a diskless database server”. An SSD drive is also attached, as a sort of control.
The SSD run takes 15 seconds. He emphasizes his use of DROPCLEANBUFFERS, runs it against scalable persistent memory, and the query takes 2 seconds. It turns out he’s running against Linux. There are sounds of surprise from the audience.
Geez, he talks fast. Bottom line: “SQL Server 2017 is 5x faster out of the box, with 50% storage costs.” <– This is about an HP product, if that wasn’t clear.
Now, automatic tuning
Bob: “Have you heard of parameter sniffing?”
Conor: “Yeah, I’ve heard of it, I think I named it”
They’re going to simulate the problem, which I won’t explain just now, but it causes stored procedure performance to tank (more or less) out of nowhere.
He’s running against a DMV, dm_db_tuning_recommendations. Someone to the left of me is actually laughing delightedly. The thing gives, as you might guess, recommendations on how to fix the query.
I believe he’s running with automatic tuning. We see the query run, start to tank, and then performance picks up when the automatic tuning kicks in.
SQL Server 2017
Bob and Conor exit, Rohan returns. After some stats, “SQL Server is now the fastest database platform in the world, period.” I look forward to a closer look at those stats.
He’s talking about R, which I have not yet played with, but about which I hear nothing but wonderful things, seriously.
- Support for graph data and queries
- Advanced machine learning with R and Python
- Native T-SQL scoring
- Adaptive query processing and automatic plan correction
There are a good many more new features…I see “management studio monthly build on the slide”…but he’s not going into them just now. I hope someone snapped a picture of that slide.
Demo – Getting started with SQL Server on Linux containers
Tobias Ternstrom (principle group program manager) and Mihaela Blendea (sr program manager) now!
Setting up a new build. Container: “a packaged application with one ore more components, and SQL Server can [now] be one of those components.” It lets you set up your local environment fast and predictably.
They’re showing how to automate the process of packaging SQL Server within your app.
She’s showing us the code for the build definition. It restores the production database, sanitizes and shrinks the database.
There’s more, but it’s not in my field and it went by FAST>
Announcing: MS SQL Operations Studio
Rohan is back, and he’s announcing SQL Operations Studio. “Free lightweight modern data operations tool for SQL Server”, which works cross platform on your database of choice. Kendra Little, next to me, is SUPER excited about this.
Also announcing special pricing (up to 30% off) for SQL Server on Linux and Red Hate Enterprise Linux.
Slide: “Optimize SQL Server workloads with Azure”. Looks like you can templatize a SQL instance…and we’re already past this again.
Cloudy cloud cloud
“Streamlining your journey to the cloud.” This is more on-message.
This is always touchy to some of us, because some of the message hints at “You don’t need DBAs/as many DBAs!” (Yes. You DO. You need DBAs for cloud databases, folks. /rant)
He’s spotlighting the path – hybrid on-prem and cloud, DB migration service, and Azure SQL Database – but the thing that’s jumping out at me is a note, “Maximize current on-premises license investments[new]”. It looks like they’re incentivizing on-prem and hybrid shops that are moving more to the cloud.
Azure SQL Database
“Focus on your business and we’ll take care of the rest.” See?
Here’s another slide I’d like a longer look at. Headlines include Intelligent DbaaS, Privacy nd trust, Seamless and compatible, Competitive TCO.
Demo with Danielle Dean
Danielle Dean, principal data scientist. “How we’re using Azure…and machine learning…”
There’s a hospital application open. Clicked on a button, and it came back instantly. It was machine learning behind the scenes apparently.
Now, ingesting tons of information into an Azure, 1.4 million rows/second.
Now looking at machine learning model.
Azure Data Factory, SSIS in Azure…
Rohan’s back, talking about the new Azure data factory.
SSIS in Azure has my attention. “Azure data factory now provides a managed environment for SSIS execution”. … “Hourly pricing with no SQL Server license required.”
Let’s see Scott Currie, the creator of Biml! Biml is another tech I hear nothing but excellence about.
We’ve got some scripts here. Announcing that for Azure Data Factory v2, full support for First Class Elements for creating Azure Data Factory assets.
Azure SQL Data Warehouse
Rohan again: “This is our flagship cloud data warehouse story.” Introducing Azure SQL Data Warehouse compute-optimized tier.
Demo: Julie Strauss, principal group PM. “Blazing fast analytics at petabyte scale with SQL Data Warehouse”
Will demo how this will significantly boost the performance of compute-intense queries. Users want fast UI, and the queries are extremely complex, and the data set is of course huge.
Query at hand has been generated by an app. Combines two analytical concepts, and so on. (In other words, this is a sufficiently complex query.) Lots of data collected, transformed, and analyzed.
Executing query, came back in 6 seconds. That’s on 100 TB of data. Not bad at all. Julie: “If that’s not impressive, you’re going to have a hard time…”
Here we have more slides, often with more information than I can absorb. Simple slides are best, kids.
Demo with Christian Wade, sr program manager. “Insights at the speed of thought with Analysis Services”.
Data is the NY taxi model. He just dragged-and-dropped a marker for 10 billion roads, and got instant BI back. That really is quite impressive. “This is clickly-clicky, draggy-droppy data analysis” for massive data sets.
That really, really was an impressive freaking demo. Wow.
Announcing: the new scale out feature is generally available, allowing you to support 100,000s of users…
You can have several read only models of that huge data set, implemented as a simple slider bar (meaning if you want 7 RO models instead of 5, just move the bar over 2 spaces). All load balanced and readily available. Cool.
Back to Rohan…
Ricardo Muti from PowerBI team. “Power BI Premium” for on-prem, cloud, or hybrid.
I’ll tell you the truth, we’re getting a bit punch from the “drinking from the firehose” feeling of this keynote. Lots of demos, which we love. This was a nice quick demo of Power BI Report Server, putting a report together quickly over a ton of data.
Back to Rohan for the goodbye.