PASS Summit 2015 Day 2 Recap - SentryOne Team Blog

PASS Summit 2015 Day 2 Recap

Here follows stream-of-thought notes about day 2 of the PASS Summit 2015.

PASS Business Update and Annual General Meeting

Adam Jorgensen, Executive Vice President and in-coming President for the association.

As a not-for-profit, the organization is required to hold an annual business meeting and recap financials for the members. The organization has increased revenue by 10%, a series of steady increases since 2008 (coincidentally, my last year as president of the organization). 2008 had 5 SQL Saturdays. Last year, there were 99 of these awesome events, with almost 1/2 occurring outside of North America! There are now more than 150,000 members, 163 member countries, and more than 86% of the countries of the world.

The organization returns about 78% of its total revenues in community activities, services, and education. The organization has reserves of 12%, still below the recommended best practice for international professional associations, but light years ahead of where we were even as recently as 2010 when I last had a seat on the board. Lots more details at PASS Governance.

The Internet of Things

PASS will return to definitely Seattle thru 2019! I'm certain that Microsoft will make the most of having our community right here in their own back yard.

Changing of the Guard at PASS Summit 2015

Denise McInerney, Vice President of Marketing.

Bill Graziano, immediate PASS President, and Amy Lewis, director at large, are rolling off this year. I'd like to point out the significance of Bill rolling off the board. Bill is the last board member to serve under the original association management company, Smith Bucklin.

The PASSion Award winner this year is … [drum roll] … Lance Harra! I can tell you from personal experience that Lance richly deserved this award. He's a tireless worker for the community and passionate volunteer who really cares about the people of PASS.

Don't forget about the board Q&A today at 3:30 PM!

And once again, the PASS Summit will return to Seattle in 2016 on Oct 26-28. (Another birthday away from home!) You can register today!

Data Management for the Internet of Things (IoT)

IoT, Edge Computing, and Cloud vs Fog Concepts

Dr. David DeWitt, Technical Fellow, Microsoft Jim Gray Systems Lab
Dr. Rimma Nehme, Principal Software Engineer, Microsoft Jim Gray Systems Lab

Rimma leads us through the IoT, distilling a very complex and somewhat abstract topic, into an easily understandable presentation. This discussion is extremely important because IoT is literally the direction of technology across the board for the majority of companies that produce physical "things", from health care to appliances to automobiles to to industry to infrastructure like sewers, water, and electricity.

The main value proposition: unconventional revenues, incremental revenues, and operational efficiency.

Four types of IoT capabilities: monitoring, control, optimization, and autonomy. We're only in the 'terrible twos' or between stage 1 and 2 for this technology.

Value to the customers is huge. Even in industry, IoT shows the "power of 1%' is billions or even trillions of dollars.

Dr. Rimma Nehme

David takes us through the "HOW" of IoT, starting with the differences between consumer and industrial IoT. For purposes of the keynote, David is focusing on the industrial side of things to highlight to "power of 1%'. Ironically, IoT is "one size fits none". Furthermore, it's entirely bespoke and must be custom-created.

IoT includes a sensor, an actuator, a connection to cloud (or another data store), and big data or a some other data engine such as a streaming database engine like StreamInsight, a feature within Microsoft SQL Server. Azure, for that matter, offers the Azure IoT Services suite of features and services to make construction your own IoT solutions. The cornerstone of this offering is Azure IoT Hub.

Growth of Internet-connected Devices

One of the really neat things you can do with IoT data is learn from it using, for example, Azure Machine Learning. This technology can have a profound impact on real-time analytics because it gives you the ability to predict outcomes. You would probably also use Azure Stream Analytics or perhaps Apache Storm on HDInsight to perform real-time query analytics and provide actionable information on data as it comes from the array of IoT devices. Note that real-time query engines watch a variety of incoming transactions over a brief and recent stream of events, say for the last 60 seconds, watching for important conditions to occur and, if they do, to take certain actions. Compared to an RDBMS, it is the queries that have durability, while the data does not require durability at all. Quite unlike the ACID-properties of transactions that we're all familiar with.

David now takes us through a real-world example using a Raspberry Pi device, a "field gateway". IoT metadata is not stored in a DBMS, hence, there is no query support. Otoh, securing the IoT devices is supremely critical. This is complicated by the fact that IoT devices must be polled regularly, while not exposing any open ports. This is hard!

But there are shortcomings in the current model. Pushing everything to the cloud has a LOT of drawbacks: storage constraints, bandwidth limitations, CPU bandwidth processing for "non-interesting" events, and more. Dewitt would like to see greater exploitation of the field gateway to make the IoT system much more robust, smart, and responsive. This reduces latency by an order of magnitude — bringing us to the concept of "edge computing", also known as the "fog". The fog is at the edge of the cloud and the field gateways. "What the fog?!?"

Dr. David Dewitt

"We are taught that you never move the data to the computation. You move the computation to the data". Rimma advocates for a shell database using a Polybase-style query engine to maintain IoT processing. Queries might be "execute once", "execute forever", "execute action", and "multi-purpose querying" behaviors. In some cases, these queries might need a new construct for OUTPUT(sync1, sync2… n) style behaviors, enabling them to access sensors directly.

Why Should a DBA / DataDev Care?

"You're either part of the steamroller or part of the road!" Let's not be part of the road. You need to get ahead of this trend because is IoT is A) still in its infancy, and B) it is experiencing explosive growth and will continue to do so in amazing ways!

The End of an Era

On a sad closing note, Rimma is moving on with her MBA and into more business assignments, while David announces his retirement and cessation of future keynotes. He gave us, the community, great praise and points out the PASS is the highlight of his professional experience at Microsoft.

P.S. Thanks to Andy Galbraith and Oliver Engels for their great photos!

Connect with me online! Facebook | Twitter | LinkedIn | Blog | SlideShare | YouTube | Google Author

Comments ( 0 )

    Leave A Comment

    Your email address will not be published.

    This site uses Akismet to reduce spam. Learn how your comment data is processed.