QUENTIN CLARK: Good morning. My name is Quentin Clark. I am the product leader for the data platform at Microsoft. I’d like to welcome you and thank you for spending the day with us today.
Data is becoming an increasingly important part of our world. And it’s emerging as a major dividend for customers and their businesses.
We’re really excited about the kinds of things we see customers doing with data and the role their products play in achieving that. You’re going to be hearing a lot more about data today.
Satya Nadella is going to come up and speak about the ambient intelligence and the need for a data culture.
Kevin Turner is going to join us and speak about the value you get from the data dividends and the kinds of value that businesses will achieve by embracing that data culture.
And then I’ll be back and I’ll be speaking about the products and showing how we achieve this.
It’s going to be a great day. I really want to thank you for spending the time, and we’ll get started in just one moment.
SATYA NADELLA: (Applause.) Good morning. Good morning and welcome. It’s a real pleasure to be back in San Francisco. I had the chance to be here, in fact, twice in the last three weeks. We first had a chance to talk about our mobile-first and cloud-first vision and focus that’s driving everything that we’re doing across Microsoft.
We talked about a couple of different aspects of our strategy. In the first event, I talked about Office 365 as a cloud service that’s going to be fundamentally transformed for a mobile-first, cloud-first world where we’re going to make Office available on every device and for everyone. And we launched Office on the iPad.
We also talked about the enterprise mobility suite at that event where we’re bringing together a comprehensive IT architecture for everything from device management to identity management to data protection.
And then last week at our developer conference at Build, we had a chance to talk about Windows in a mobile-first, cloud-first world. We started off talking about the universal Windows applications, which really increases the developer opportunity across the entire family of Windows, from large screens to small screens, and having that commonality of API surface area, which we think is a big differentiator for how Windows and our tooling around Windows creates the developer opportunity.
We also had a new update to Windows Phone. In fact, the reviews of it I think came out yesterday and they’re coming out even still. We introduced this new feature which is this intelligent agent in Cortana, which is just an exciting thing. It’s got a lot to do with, in fact, today’s topic even.
And then we talked about 250-odd features. Scott Guthrie talked about 250-odd features of our Azure product and service, which is the core cloud infrastructure for developers and enterprises. So we had all of that go on in the last couple of weeks.
And today, I want to talk about another aspect of our strategy as we move forward in this cloud-first, mobile-first world. And that’s data. And more specifically, the data platform. It’s very central to our vision going forward. It’s very central to our customers. So we want to spend the time today talking about how we’re moving that agenda forward.
Now, one of the things is it’s already a pretty big business for us. It’s one of those businesses that’s hidden. It’s just one of those things that we don’t like to talk about it as much. But it is a pretty significant business. SQL Server alone is over $5 billion for us, and it’s growing at a rapid rate.
We have new growth in Azure storage that’s at unprecedented levels. In fact, it just doubles every six months.
We have, of course, Excel on the other end, which is the most ubiquitous tool when it comes to people dealing with data. So we have a pretty broad set of technologies that come together to make our data platform.
But of course what’s most important to us and to our customers is: How are we evolving our data platform going forward in a world that’s mobile first, cloud first? And that’s what we want to talk about today.
When we characterize the future as we see it, the world view we come to is this notion of ubiquitous computing and ambient intelligence.
There are three trends that are at play at increasing levels. The first one is the ubiquity of the computing fabric itself. It comes in the form of sensors, it comes in the form of devices everywhere — big screens in the living room, big screens in the conference room. It’s that core evolution of silicon, hardware and software that just puts computing pretty much everywhere a human being is present, where humans are interacting with machines.
The second trend that is also pervasive is the increasing rates of digitizing every experience that you have with computers. So, for example, we have every human interaction with other humans just completely digitized. The interactions between humans and machines is increasingly digitized, as well as the interactions between machines. So these are all trends that are accelerating.
The last thing, of course, is that you have now this enormous capacity to reason over all of this digitized information to, in fact, enhance those very interactions and experiences that are being digitized. So these three trends together is what form the world of ubiquitous computing and ambient intelligence.
And if you look at that world through the lens of data, what you see is this constant, ever-increasing data exhaust that comes from ubiquitous computing. You can envision this as logs from our servers; it can be data from sensors; it can be you social stream data, which is unstructured information. It can even be transactional information that’s happening at an increasing rate. So you have this constant exhaust. And of course, that all acts as the fuel for ambient intelligence.
Now, the question is: How does that magic happen? How do we go from data exhaust to this notion of ambient intelligence? And that’s where you need to build out a platform and you need to take a platform approach to be able to bring what have been disparate sets of technologies together. And that is, simply put, our vision with what we describe as the data platform or the ambient intelligence platform.
We want to take an architectural approach that brings together different products from Excel on one end to SQL Server and Hadoop on the other end. So these are different products that all have to come together in one cohesive architecture to create this notion of ambient intelligence.
We also want to bring together technologies. Things like in-memory, which really improve the speed and throughput of everything that you do around data as part of that architecture.
You want to be able to bring new capabilities like machine learning that improve our ability to reason using statistical methods that we’ve not done in the past to this very same architecture.
So that’s really what we want to talk about. And so if you click down on this notion of ambient intelligence, there are three core things. And it starts for us with people. It starts with everyone in an organization having the curiosity, having the questions, trying to test out hypotheses, trying to gain insights and then take action.
So to us, any notion of building out a data platform or an ambient intelligence platform has to start by weaving people into the center of it all. And the way we do that is by really taking the Office canvas and making it data aware.
You can sort of say PivotTables were the first generation of it. But now we’re going beyond that by making it possible for you to have, inside of a PowerPoint presentation or an Excel spreadsheet, access to all the data all the time and for you to be able to ask simple questions and get answers. Not because we’ve cut off the data into some silo that goes with an Office document, but think of Office as, in fact, the canvas or the surface area or the scaffolding from which you can access all the data, do the complex queries, but in a very natural way.
So to that, if there is one core differentiating aspect to how we approach the problem of data, it’s perhaps that. It’s to transform Office as the UI for data.
The second piece is also to transform the core middle tier of data, which is the analysis and processing layers, so that you can deal with the variety of data that we’re faced with. It’s the structured information coming from your transactional system. It’s the unstructured that’s coming from, perhaps, your social streams. It’s the log data that’s coming from your servers.
You want to be able to reason over all of that. That means you want to bring together what has been, to date, the two different worlds of power data warehouses built on SQL technology, as well as technologies like Hadoop into one fabric on which you can do analysis and computation.
This is a place where we will also bring in new capabilities around machine learning over time so that you have the same technology fabric for you to be able to do analysis.
The last aspect is around data itself and the storage system itself. And now, in this area, there’s some fundamental changes happening, starting with just the memory hierarchy. If you think about it, the late Jim Gray used to talk about how the disk is the new tape. And guess what, now it’s true. Now you have in-memory technologies fundamentally getting plumbed into the full transactional systems, BI systems, as well as the data warehousing systems of SQL that make the throughput and speed of these data-processing systems so much more faster.
So that fundamental change pretty much changes all aspects of how we thought about data and how we thought about the speed with which we did business around data.
So these three areas of people, analytics and data storage is what constitutes the build-out of the ambient intelligence platform.
Now, talking about technology is one thing, but if you really as an organization want to change anything, it’s about culture. And when it comes to data and to be able to truly benefit from this platform, you need to have a data culture inside of your organization.
For me, this is perhaps the most paramount thing inside of Microsoft. If you look at it, what it is that we are in the business of? We create services and products and devices. And the thing that we need to do a better job on than anyone else is to be able to learn from our customers and the data, the exhaustive usage, and to continuously improve our products and services. That’s our job No. 1, by far.
And it’s not going to happen without having that data culture where every engineer, every day, is looking at the usage data, learning from that usage data, questioning what new things to test out with our products and being on that improvement cycle, which is the life blood of Microsoft.
Now, it’s not just about engineering and development at Microsoft; in fact, every aspect of Microsoft from finance, how they’re going to do continuous forecasting in this very rapidly evolving world of ours. How does HR do people analytics? How does real estate, in fact, manage the real estate facilities? Because now if you’re in the cloud business, real estate is not just about housing people, it’s more important to us given all the datacenters that we run and the environment around the datacenters because that’s basically tied to our SLAs to our customers.
So there is every aspect of Microsoft business is being fundamentally transformed because of data. And that doesn’t happen because of technology, it happens because you have to build deeply into the fabric of the company a culture that thrives on data.
So we want to talk about this. And we want to show you how Microsoft itself is using data and inculcating this notion of data culture across the various aspects of how we run our business. And to show you that, I wanted to invite up on stage Eron Kelly from our data platform team. Eron?
ERON KELLY: Thanks, Satya. (Applause.)
Well, it’s great to be here and talk about how Microsoft employees have fully embraced the data culture. And we’ve done this by transforming the tools that you use every day like Office and really allow them to easily unlock the value of ambient intelligence.
No longer are we looking at static reports and charts. Now it’s about interactive and rich experiences.
Let me show you a couple of vignettes. I’m going to start with our cybercrime team. Now, at Microsoft, we have a cybercrime team whose mission is to identify botnets when they enter the wild, understand how they work and analyze them, then shut them down and help local law enforcement arrest the criminals.
Now, of course, as you can imagine, that generates a lot of data. And you need to analyze a lot of data. But the challenge you have is the lawyers and investigators aren’t data scientists by trade. And so when they see data like this, it’s interesting and important, but it’s hard for them to really see the big trends.
That’s where a tool like Power Map comes in, which is part of Excel. It allows me to now richly visualize that data in a 3-D interactive map. It’s kind of like going from the “Andy Griffith Show” to “NCIS” if you think about it. (Laughter.)
OK, so what you see here is a list of all these infected machines around the world. And one of the things we’ve learned over the last couple of years is that these criminals don’t like to have viruses installed on machines in their home countries. Why? They don’t like the idea of local law enforcement coming after them, right? They’d rather be at arm’s length.
And so if we look at this map here, we can see that there’s a lot of infection here in the U.S. So the criminals are probably not in the U.S.
So if I span the globe here. Let me go check out Western Europe. Oh, Western Europe, a lot of infection there. Those are probably not where the criminals are hiding. But, gosh, there is a very bright line there that separates Western Europe from Eastern Europe, specifically the Ukraine and Russia, right? OK, well, in fact, as our engineers started to reverse engineer the botnets, they saw that they were designed to not install on machines with Ukrainian and Russian language packs.
So armed with rich information like this, the investigation team was able to narrow their investigation and rapidly catch the criminals. It’s a great example of how you’re accelerating that insight to quicker arrests and richer experiences. So here’s just one great example.
So let me switch to finance. Satya mentioned our finance teams. And our finance teams have been using these tools for all kinds of things, including the incredibly exciting area of cost management.
Now, just at the mention of cost management, I see that the gentleman in the front row is already starting to yawn, and that’s OK, I understand that. Because this is kind of how we would do it, right?
So if we looked at airfare at Microsoft, we spend hundreds of millions of dollars a year in travel. And so it’s an important area to focus on. But, historically, it was like this, right? You go through slides talking about costs. And, yes, the gentleman in the front row has now completely fallen asleep. And I don’t blame him because at Microsoft, we would do 27 slides per country, times 50 countries, 1,350 slides around airfare and cost management. Really, really difficult.
Now, it’s just one Power View. Interactive and rich, one slide versus 1,300. I can choose a country like France. I can see how the cost in France over the years has gone up. Maybe that’s a bad thing. But if I go over here and look at what percentage of trips are external versus internal, I can see that my external trips are going up. That’s a good thing. That means my sellers are out visiting customers, creating better experience with them.
If I switch and look at internal, I can see that my internal costs are going down. So the number of trips that people are doing to visit internal customers and visit internal folks is going way down. That’s a great thing. And it’s because, in France, they’ve used Lync a lot.
Now, one of the really important metrics, of course, is cost per mile. And you can see cost per mile has been going down over the last couple years. What are the big drivers? Well, of course, one of them is when you schedule your ticket. If you schedule your ticket over 21 days out, you can see that the cost per mile is cut in half. If you’re late, like me sometimes, you can see how the cost per mile is more than double that average.
OK, that’s intuitive. People understand that. But in the past, that was buried on slide 17 — it was hard to see. Now it’s front and center. Everyone can interact with it and see the data. It’s very, very rich. It’s a great example of a data culture.
Now, speaking of everyone, Satya mentioned the real estate team. And our facilities folks have been using this technology as well to better understand power consumption around the buildings within the Redmond area. And so if I take you to some of their data, again, it’s kind of like this. It’s hard for sort of the non-data scientists to really find the insights. And, again, I’m going to go to Power Map, which is a great way of visualizing data.
Now, in this case, I’m looking at two things. I’m looking at the power consumption by building, that’s the bars. And I’m also seeing what the average temperature is in the Redmond area, that’s this heat map. And what’s really cool about this is I can just play this animation and see the interaction.
Now, no, this is not the bass line for the latest Macklemore song. This is, in fact, the interaction between power goes up as the temperature goes down. As the temperature in Redmond goes up, the power consumption goes down. So you can see that interaction. It’s very rich, it’s very intuitive and it’s engaging. Everyone can find an insight from that.
But I want to drill deeper. So I want to come over here and talk about an incredible new feature as part of our Power BI service called Q & A. Now, what Q & A does, like it sounds, it allows you to ask a question and get an answer.
So if I start here with that same data set, let me just take a look at, say, total energy by date. So now what I’m doing is I’m looking at the total energy by date. If I can type. Yeah. Total energy by date.
And I can see that same pattern that we saw in the past with — there we go — you can see the same pattern that we saw on the animated Power Map. You can see energy consumption is much higher during the winter months, and it’s much lower in the summer months. Unfortunately, in Redmond, we don’t have a lot of need for air conditioning in the summer. I wish that wasn’t the case. But you can see that consumption. Rich, interactive chart.
I can also, say, look at it by building. And what Q & A does is it says, “Hey, you’re interested in a comparison of buildings, let me give you a bar chart.” And I can immediately see that Braeburn 1 is using a lot of energy.
All right, so we need to go out and talk to the folks at Braeburn 1. Or do we? Maybe if I look at another comparison and if I compare it versus people, I can see the scatter plot. And now I see that Braeburn 1 actually has one of the largest buildings on campus. It has more people annually that go to it than other buildings. And so the energy consumption is kind of on that diagonal like I’d expect, whereas the Commons is really standing out.
Now, for those of you who have been to Redmond, you’ll know that the Commons is kind of a retail space, there’s a lot of restaurants there, and so it’s kind of natural that there’s a lot of energy consumption there. And there’s not a lot of full-time employees.
But Building 121 is definitely standing out. Building 121 is definitely an anomaly. So let me go drill into that and look at Building 121 versus, say, Building 34. And Building 34 has a similar size. I’m going to look at that by date. And it’s a similar size. And now I’m really finding an interesting insight.
During the summer months, the energy consumption is about the same. But in winter, you see Building 121 is using a lot more energy. Well, it turns out that someone was running a furnace in Building 121 heating up the garage to the tune of $56,000 a month. Now, historically, the facilities team was not able to see this because they just assumed, hey, that’s the normal pattern for Building 121. When, in fact, once they were able to make this relative comparison, the true insight came out. And what’s great about this is it’s simply asking a question and getting an answer. It’s very interactive.
Now, the last vignette I want to show you is around marketing. And at Microsoft, Bing is this amazing asset for us. It’s an incredible corpus of ambient intelligence. And one of the cool things with Power BI is I can create these questions that are kind of ready to go. So I can just click on it and it asks the question.
In this case, I’m looking at Bing searches for phones by market and category. And the idea was, hey, if we look at what people are searching for, what category of phones and where, can we better tailor our marketing and advertising? And so you can see on this map, hey, here’s where low-end phones are being searched for. Here’s where high-end phones are going to be searched for. Very, very interesting and intuitive to help me with my marketing.
But I may also want to understand when people are doing the marketing so I can better tune that in. So if I switch that today, I can start to see an interesting pattern, right? People are searching for phones during certain parts of the day. And maybe if I want to look at that by day of week, I’m now starting to see a pattern. OK, Wednesdays and Thursdays, people are searching. Saturdays and Sundays, not so much.
What if I compare this to, say, phone activations? And instead of looking at it as a scatter plot, I want to look at it as a line chart. Now the real insight comes out. I can see how people are searching for our phones on Wednesdays and Thursdays, but they’re buying and activating over the weekend.
So armed with that information, as well as an understanding of where people are searching from, I can better tailor my advertising to focus on Wednesday during the week, that’s when people are making the decisions, and then make sure that the right messages are landing in the right locations.
So a great example of how we’re using the data culture at Microsoft to transform our business and really unlock the power of ambient intelligence. Thank you very much. Welcome, Satya. (Applause.)
SATYA NADELLA: Thank you, Eron. Hopefully, that gave you a good feel for how we at Microsoft are using this technology platform to change, fundamentally, the culture and how people approach their work. And as I said, it starts with that curiosity and questions that each one of us has.
And the key technology breakthrough for us is how do we make it possible to use the tools that you use on a daily basis? Which is Office. And, yes, make the power of data just shine through. In fact, there’s some very amazing technology behind that PowerPoint presentation that had access to all that real estate data. There is a complex columnstore, which is powering all of that. So this is not about just some data in an Excel spreadsheet or a PowerPoint that goes with the PowerPoint or the spreadsheet. It is, in fact, a cloud service in this case that has all the data that’s available to the user all the time.
But more important than that is how did this all come about? And, perhaps, that’s the story that we want to evangelize. That’s the story that we want to learn from ourselves and our customers’ experiences and really get on this improvement cycle.
If you look out at — even that real estate example required IT development and end users to come together to create that. For example, there are over 2 million sensors inside of Microsoft, this is in the real estate facility. That’s just generating close to 500 million transactions per day. That’s all getting collected in a Hadoop system. Then a developer goes to work on it and reduces that using MapReduce into something that then can be loaded into a SQL analysis server, so that means that’s possible for you, then, as an end user to go in with something like Excel and create that semantic model that powers the Q & A feature.
So the fact that at the end of the day, the end user wants to just be able to go to a SharePoint site and start asking questions. But behind that, you really had IT build the systems that could scale for the throughput required. You had developers work the entirety of the processing stream so that you could, in fact, take all that data and make it into a warehouse which then enabled a power user in Excel to be able to create a semantic model, which then enabled everyone inside of the organization to be able to have access to that Q & A feature.
That notion of being able to empower individuals or teams and entire organizations, that was done by the coming together of IT development and end users, is a sensibility that we have in many other areas. And we want to bring that to the world of data. That means you have to take a first-class job of tooling. So even for developers, how do we take all of the developer APIs from Hadoop to SQL to be able to do things like sentiment analysis on streams of data and through ML techniques? All of that, how do we bring it together with Visual Studio is a very important problem, and we want to bring our best work there.
How do we then have the systems which deal with a variety of different data, from the NoSQL databases to the SQL databases, to be able to do different kinds of processing, like MapReduce or real-time streaming? So those are the aspects of uniqueness that we bring, but bring it together with a cohesiveness of an architecture that brings all these things together.
Now, the milestone for us today is that we have three new products that we’re announcing that make up part of our data platform. The first one is SQL Server 2014. This is a pretty breakthrough product for us in the industry. For the very first time, we have now in-memory technology built into every workload of SQL Server. So that means there is online transaction processing, or OLTP, now can take advantage of in-memory without changing your existing applications, and that’s a very key point for us.
Second, it’s about the BI workload that has that compressed columnstore in memory that accelerates your ability to get analysis out of your system. And to be able to build out your data warehouses using in-memory technology. So we now have the technology in SQL Server that accelerates every workload by significant throughput, and we’ll talk more about that.
The second service or the second product that we’re announcing today or launching today is the analytics platform system. Now, this, for the very first time, brings together the parallel data warehouse of SQL with a Hadoop region. So that means now you have the ability in one affordable appliance to be able to do queries across both of these.
I mean, you can imagine the scenarios. You can imagine having data from your transactional system, your log information from your servers and websites, as well as social streams and the ability now to be able to query across all of this. There’s a piece of technology that David DeWitt and his team were fundamental contributors to called PolyBase, which is just amazing in terms of bringing the power of SQL language now to be able to span across Hadoop as well as regular SQL. And we think that that’s some breakthrough work, and that’s all part of this product.
The last thing we’re also announcing today is a preview of a new IOT, or Internet of Things, service. It’s called the Azure Intelligent Systems Service. This makes it possible for you now to be able to collect data from all the sensors and servers, bringing that into the cloud so that you can take advantage of the rest of the data platform to be able to do the analysis, the machine learning, and so that is a service that I think will take out all of the friction that exists in being able to connect the cloud with the Internet of Things trend that is only going to increase. So we have an exciting set of customers and third-party ISVs who are already working with the service and learning from it. And so we’re very excited to have the preview launch today.
So those are the three specific announcements. But we’ve been at work on a rapid pace because these three products join this full platform that we have around data. So as I said, it starts with the core capabilities of Office itself, with Excel, SharePoint and the Power BI service, which we launched just around six weeks ago.
In Azure, we have all the storage systems and processing systems, from HDInsight, which is our Hadoop service, to the Azure databases, Azure storage, so these are the capabilities that are available in the cloud.
But one of the other interesting aspects of Azure is every workload of SQL runs in the infrastructure as a service. In fact, it even gets better because you can take SQL 2014 and tier it with Azure. So in order to get something like high availability, which is a very, very important aspect for IT, you can in fact get that from Azure because you can have read-only secondaries and high-available configurations of SQL automatically tiered for you from your on-premises deployment.
So we’re really not even thinking of these worlds of cloud and on-premises, private clouds as two different and distinct things. But it’s one integrated computing and storage fabric for customers.
Lastly, we have SQL 2014 as well as the analytics platform system. So that describes the breadth of our data platform. But the thing that’s most important for us is the data culture. Because, in some sense, a lot of the technology that is there today and how we will evolve it is going to feed off of the feedback cycle. In some sense, we will feed off of the data that we get on what is the friction? How are people using these products to bring about that fundamental change with an organization so that they can get the data dividend? Because one of the things that is most latent in terms of value inside of an organization is the value that’s contained in the data.
And the question is: How do you empower the people inside of the organization to be able to take full advantage of that and really reap the rewards?
And so to talk more about that, I wanted to invite up on stage our chief operating officer, Kevin Turner, and have him take you through a lot of what our customers are doing in terms of data culture. Kevin? (Applause.)