THE SQL Server Blog Spot on the Web

Welcome to SQLblog.com - The SQL Server blog spot on the web Sign in | |
in Search

Louis Davidson

After PASS Summit 2016 Recap (As seen by me!)

In my last blog entry, I promised to blog about the PASS Summit each night when I got back to the room. This was a failure for two reasons. 1. I was always out at night and then exhausted. 2. I forgot the keyboard to my Surface Pro. I tweeted about it, and was picked on by the @surface twitter account:

image

But I did tweet about the event rather frequently, as it is much easier to capture your ideas and comments in 140 character spurts (and favoriting other posts saves typing too.) If you want to read all of the tweets about the summit, look for the #sqlsummit hashtag on Twitter.

The first day was the Microsoft Keynote. It was led by Joseph Sirosh and while there wasn't a tremendous amount of stuff that directly excites this relational engine programmer (though love was shown for how awesome the SQL Server Engine is, both on prem and in the cloud), some of the stuff that was shown was really cool:

1. Showing the various sources of data you can use with Polybase

2. SQL Server on Linux - Not that I will ever use this, but it could be a boon to SQL Server usage as time passes (and for a relational programmer, you would not really notice much of a change anyhow)

3. Azure Analysis Services is coming soon

4. Azure SQL DW has had some tools created to make it easier to get started with (RedGate has a free tool at http://www.red-gate.com/products/azure-development/data-platform-studio/), and as Tim Ford tweets here: (https://twitter.com/sqlagentman/status/791316930703478784), you can get a free month of SQL DW to give it a try.

The biggest takeaway was just how much data is going to affect our lives as time passes. Last year, my reaction was that the keynote was a bit creepy, taking mapping DNA and predicting health. This year, it was a couple of examples that were really cool, including some apps, websites, a few game examples, and sentiment analysis of the book War and Peace (https://twitter.com/drsql/status/791320039303491584) by Julie Koesmarno.

An interesting turn of technology was the push towards "intelligence database" platforms. Something that many of my colleagues have discussed for years has been to leverage the data tier to get work done faster, and more reliably. What had always been missing in those scenarios has been scaling out. Hence we were constantly limited to how much we could do on a single computer. Two things have changed since those days. 1. A single computer can do as much work as most organizations need to. 2. Scaling out is far easier when dealing with read intensive scenarios. There was a demo of SQL Server 2016 handling millions of transactions where the reality was orders of magnitude lighter (and we are talking fraud detection for major credit card companies).

However, the most moving demo finished out the keynote, and it was the closest to creeped out that I got. There was a computer guessing ages, (I think) gender, etc. Then the computer was describing the surroundings. The the computer was reading a menu at a restaurant. And then you realize this was a computer helping a blind man. Wow. That was just an amazing use of technology.

If you want to know what Joseph Sirosh (the Corp VP for the Data Group at Microsoft) felt were the top five announcements, he shared it here: https://twitter.com/josephsirosh/status/790950683138596865. Stuff I didn't mention was really outside of what I know (ok, I admit it, care) about (I do only have so much time!)

-------------

After this I attended several pretty great sessions:

  • Why Datatype Choice Matters from Andy Yun, where he covered some of the internals of datatypes. The best part for me was the statement that "NULL isn't a value, it is a state of being unknown, undefined.  Hence the null bitmap in the physical record of a row." While I have written about NULL probably hundreds of times, it is good to be reminded of this point, that NULL isn't really a value, even though it does feel like it.
  • Building an SSRS monitoring system with Stacia Varga (a cowriter on this book). She covered a lot of stuff about logging that I may never use, but one thing I learned about that I might directly use is logman.exe, which lets you capture perfmon counters. There is an article here about capturing SSRS statistics: https://msdn.microsoft.com/en-us/library/ms159809.aspx).
  • Then Tom LaRock and Karen Lopez duked it out again talking about time bombs you have lurking in your database code. You know like NULLs no one understands, identity column values that no one pays attention to when the values run out.

----------------

Something I am keen to learn more about came in two sessions: Buck Woody the first day and Dr Cecilia Aragon. Data Science. I don't know if I could, or would want to, become a data scientist. But in either case it leads me down the path of wanting to make sure that databases I create are ready to be a source of some of that data. I have always been a proponent of tailoring my OLTP database designs to capturing every detail that is possible. For example, cause an effect, when it is direct (such as a shipment to an order), or indirect, (such as a follow-on order that the customer tells you, or gets in a link to, a previous order.)  Data Science is about learning more about everything, and the more answers you can provide an algorithm, that can only help you see others behaving the same way.  Capturing additional data that isn't needed immediately is not always something that is greeted by developers with a hearty smile, but it is almost always going to be useful.

Buck Woody pointed out a website (http://tylervigen.com/spurious-correlations) that has some excellent, messed up, correlations that you can make using data. Such as "Per capita consumption of chicken" and "Total US crude oil imports':

image

I eat a lot of hot chicken to try to help, but I am only one person!  These correlation were highlighted even more by Dr Aragon, who had a couple of very interesting quotes that piqued my interest:

"Data science is driven more by intellectual ecosystems and software ecosystems than by hardware"

(Paraphrasing)"Humans not gaining exponentially greater cognitive capacity. "

"Big data is data that is 2 orders of magnitude greater than you are accustomed to"

For me, these three quotes really put Data Science in perspective. People are now, and have been, very intelligent, regardless of how things seem at times. But what we really lack is the ability to process concepts quickly. People make algorithms, and could slog through data manually, but rather let computers whip through data and give us decisions. Will there ever be a time where machines make correlations that are completely wrong, but they act on them anyhow? It reminds me of Robot Santa Claus on Futurama who judged everyone naughty, the person who was naughty, and the person who told on the person.

Will we ever make a machine that can come up with algorithms, and understand what is a meaningful correlation without some human logic? Heaven knows that every person who creates a machine won't be good at heart, but could machines ever be machines without people?

It does all remind me of the Pete Townshend song "Man and Machines" from the Iron Man album :

"Man makes machines
To man the machines
That make the machines
That make the machines
Make a machine
To make a machine
And man and machine
Will make a machine
To break the machines
That make the machines..."

On Singlularity Hub I was reading an article about the subject of AI, while it isn't the same thing exactly, has many of the same problems. There was a statement:

"Based on deep neural nets, the AI impressively mastered nostalgic favorites such as Space Invaders and Pong without needing any explicit programming — it simply learned through millions of examples."

If you stop at "without needing any explicit programming", this sounds pretty creepy. But if you give the computer an example of a successful solution, perhaps even millions of them, and combine this with the fact that computers don't make tiny mistakes (you know, what makes games fun!) it isn't that the computer can learn by itself. Just that it can try, fail, adjust, and repeat a LOT faster than people. But it still takes a human to guide the process.

-----------------

The second keynote had two major parts. First was the PASS business stuff. We have more chapters, more members and want orders of magnitude more people. One way of pushing this direction is, much like the MVP program did, including the entire data platform. PASS no longer means Professional Association of SQL Server, but just PASS. New logo too:

image

The little symbols represent what PASS encompasses in who PASS is as an organization, and we as PASS members. Interesting enough, but I always worry that things are going to go sideways and we will end up in a different community of mostly the same people. Time will tell.

The second part was an excellent keynote address by Dr David Dewitte. It had some interesting details and comparisons of online data warehouse products, but was a lot broader than that. Good overview of internal stuff that can only help your career. I won't cover anything about it, go here (http://www.sqlpass.org/summit/2016/PASStv.aspx) and watch it.  Best quote for me: "SQL Server, Best Query Engine". But other companies are doing great stuff too.

----------------

Then I went to a discussion about the way sessions are chosen. PASS choses sessions in a very interesting way, but really I think they do a good job. No matter how you do it, someone's feelings will get hurt unless you use my favorite method for SQL Saturday session choosing. Everyone gets a session if you haven't angered the organizers in some meaningful manner. Best way to anger the organizers: don't show up without a good excuse. Yes, it happens too often. And that, along with harassing others at an event (or multiple events), is something that takes a while to get over. Best way, be apologetic, attend events and don't be a jerk again.

-----------------

The other big thing that happens on the second day is that a group of folks wears kilts to PASS to show support for Women in Technology. This year, I was one of those people. It was not a normal thing for me, and not something I expect to do for a SQL Saturday unless for something special. Want to see the picture. Click this link to see Jamie Wick's tweet of a picture that was taken: https://twitter.com/Jamie_Wick/status/791739875439456256

-----------------

Friday, we woke up to a rather interesting sight for a PASS conference, even more interesting than myself in a kilt. The sun came out:

Attended one more regular session of note: Tim Mitchell's Deep Dive of the SSISDB catalog, where I knew most everything, but using Data Taps to capture intermediate results like you might to a temp table in a SQL Query Batch was very nice. I hope to run a series of blogs about some work I have done with the SSISDB catalog over the next year or so. Another interesting idea, using SSISDB versions for regression testing. Run once, deploy new, run again, compare results, then revert.

The other thing I went to was Speaker Idol, supporting my man Robert @SQLCowbell Verell. We co-lead the Nashville SQL User Group, and it helps us if Robert gets a speaking slot :) Robert was wild-card/runner up of the day (there are three rounds of four, with a final round of four to complete the day), and he did a great job. I really felt nervous for all of the people who participated, because what pressure. I have long sweated the process of speaking, because all of those eyes staring at you, seemingly expecting perfection (actually just expecting to learn a few new bits and pieces.) And here, while you have 10 eye staring at you, this time actually expecting perfection. In the end he didn't win, but he certainly didn't embarass himself, since he made the finals despite having a yellow background for text in SSMS that still is burned into my eyes.

------------------

Then it just sort of ended… No fanfare, just a walk down to the Moore Theatre to catch Ian Anderson do his Jethro Tull rock opera. I hadn't even noticed there being concerts I cared about in the area, and prior to this year I would have never wandered that far from the hotel most nights, but I discovered the ease of Uber while there, which made walking less scary, since I occasionally aggravate my knee when walking as much as I did this week!) While there I ran into Rodney Kidd, who had a lot of great stories about music, walking back from the show (we were both at the Homewood Suites.) Add that to the stories that Pat Phelan shared at breakfast that morning about his cool experiences, and I had a great time even outside of the event.

Well, can't wait until next year!

Published Wednesday, November 02, 2016 4:17 PM by drsql

Comment Notification

If you would like to receive an email when updates are made to this post, please register here

Subscribe to this post's comments using RSS

Comments

 

email database marketing said:

Its an awesome bit of learning , Thanks for sharing the critical and important data

with us , i got some new thoughts from it . You're the best to share with us .

Much obliged! http://www.latestdatabase.com/

November 3, 2016 6:10 AM

Leave a Comment

(required) 
(required) 
Submit

This Blog

Syndication

Links to my other sites

Archives

Privacy Statement