Modernization Hub

Modernization and Improvement
Innovate with Modern BI in the enterprise

Innovate with Modern BI in the enterprise

Good afternoon, everyone. Thanks for coming out here today to
the Innovate with Microsoft BI for the enterprise. My name is Muhammad Ali. I work inside the product team for business platform applications
customer success team. And what I hope to do is
actually implement BI solutions with enterprises and customers. And so today what I really wanna
to start off with is something you probably all seen before is
this notion of data culture. And I think in the last ten years,
customers and enterprises have gone a long way into realizing how
important data is for them. And this I think would be pivotal
in the next ten years as well with AI and machine learning and all those nice things about
bots taking over the world. And what I really wanna spend
the time today is talking about what I modern BI inside
the enterprise and what it may mean in the feature as
well from a Microsoft standpoint. And so
I may say some controversial things, hopefully they won’t be too
controversial for today. And so, you may have seen this
slide presented many times before. The three wave of motion
of BI in the Enterprise. About ten years ago when you first
were looking to do a BI solution, you’d have to get IT involved and you’re really looking at
a year timeline minimum. Just to get that implemented and
resolved and getting your business
users to use it. The second wave then came along is
making Excel more easier to use for analysts and
to drive their own data insights. And so analysts would try to
avoid IT as much as possible but there were still some areas
where they required IT and IT had to take their time. The third wave is where
Power BI came in, is this notion of
self-service analytics to empower everyone inside an
organization to carry out analytics. And that’s where
the third wave came from. And I think the third wave and
fourth wave, we’re kinda saying a lot more now,
is this notion of BI is great but it wouldn’t be great if you could
integrate BI inside an application and allow me to act on the business
insights that you showed me. And that’s where I see the fourth
wave really taking place, is this notion of artificial intense
machine learning, playing a very pivotal role in the way BI will
head forward in the future. And so what I really wanna spend
today’s session on, is the agenda for today is actually split up
BI into four succinct boxes. And the first one I’m gonna spend time today is actually go through
the product stack of Microsoft BI. And that normally will revolve
around two main products, Power BI and SSRS. And I’m gonna go through a lot
of demos in Power BI and SSRS and actually show you a lot of
the demos mainly centered on those product today. The next section is scale out,
how do you scale out Power BI and SSRS when you’re dealing with
sheer volumes of data, millions, hundred millions, terabytes of data,
petabytes of data. And so I’m gonna go through
a SQL Server Analysis Services as a service in Azure which released
a couple of months back, and actually go through
Azure Data Link and show you when you would
use one versus the other. And I’ll actually show you how they
complement each other in the real world enterprise. The third section is something that
you guys have probably heard a lot about, is this notion of advanced
analytics, machine learning, AI. Cognitive services and I’m just
gonna spend a brief portion of the presentation and actually go
through what that means and how you can integrate them inside your
business intelligence applications. And then the fourth one is where
I see the fourth wave going for BI, is how do you amplify, how do
you make BI actually actionable. How do you actually act on the
insights that you see inside most analytics, if you look at a graph,
what do you actually do from that? And so, there’s a lot of work going
inside Microsoft to actually make the fourth box a reality. We’ve come a long way. And so, we started off with
something called Power BI embedded. And I’m just gonna quickly,
briefly spend some, about ten minutes actually going
through what it means to be at the. As always I wanna make this as
interactive as possible, if you guys have any questions, please feel
free to ask during the the session. I’ve got enough material to
last me one and a half hour, but I’ve set aside about
15 minutes for Q&A. So please feel free to just
pop up over your hand and I’ll happily answer. Yeah, I’ll start off with
the first box, Power BI. How many of you have actually
used Power BI today? So quite a handful. How many of you actually
implemented Power BI in Enterprise? So not as many as the first,
but quite a few of you. So Power BI has come a long way. I think it’s coming to its
second-year anniversary of GA. And Power BI has really served the
promise to help users self-serve. I want everyone in my organization
to be able to get insights to their data and actually make
good data-driven decisions. And so
Power BI comes in three flavors. There’s a Power BI desktop tool,
which is completely free for everyone to use, and it’s free
as in there’s no catch to it and you can go download that today. Then there’s a Power BI service. What you can, once you’ve driven and
find your insights, found your reports that you wanna create and
offer, you can actually then publish them into Power BI Service and then
share them within the organization. And then there’s Power BI Platform,
which is, I wanna make an app. I wanna satisfy myself,
I wanna make insights and then deliver them to people
outside my organization. Customers of mine so that I can then
make potentially a good business model and actually setting
good analytics to these guys. You guys may already notice but
Power BI is something worth demoing. So instead of you going for the
features one by one, I’m actually gonna just spend the time, actually
demo it, live in front of you guys. So, I’m just gonna come
of out my presentation. I’m gonna minimize it. I’m just gonna to open up
Power BI Desktop for the first time. As you can see, I’m a huge
supporter of Roger Federer. Hence the picture there. And this is just gonna
load up Power BI Desktop. And when you first come
into Power BI Desktop, it takes a couple
of seconds to load. You can sign in, you can see some
recent articles have come up and we release updates for
Power BI Desktop once every month. So you can expect a handful of
features to come up once every month and normally they
are driven by you guys. We have something called the user
voice where you can submit ideas. We take the top ten ideas of that
list and try to implement them as fast as we can and bring them to
the product as fast as we can. I say is that’s very unique for
Microsoft, to release once every month for
a product. And it’s becoming more and
more mature. And we’re still trying to keep up with that same velocity
that we’ve originally had. And so I’m just gonna
close this window down. And for this, I decided to actually
just go to a web URL this morning. And see if I could just
extract the data sets. So I was very interested
in olympic data set. And so I was just gonna
extract the olympic dataset. So the first thing that you probably
wanna do when you go into PowerBI is click Get Data. And this is where you wanna bring
in data from a wide variety of resources. And as you can see there’s
a long list of this, and we’re constantly adding to this. We’ve added support for SAP in the
recent months, we’ve got in-parlor, we’ve got Azure Data Lake
dded in here as well. And you see the list goes on. I think we’ve got over
50 connectors, and we’re constantly adding to this. Today for the demo what I’m going to
do is actually connect to something called web-end point, and I’m just
gonna go to the other connector. I could use R as well,
if I really wanted to. And I’m just gonna click Connect,
but before I do that, I’m actually just gonna pull out
the link that I need to connect to. And actually show you what
this link looks like. So, I’m just gonna open
up my trusty browser and let’s see where this goes. And so this is gonna take
a couple of seconds to load. How many of you have ever
cleansed XML data before? And realized what a pain that is? Yeah? So, cleaning this XML data normally
would require me to go into a language like Python, C#, spend
some time, try to really understand what the schema looks like and
actually go ahead and cleanse it and then provide it in a form that
I could actually analyze. Inside Power BI, we have a tool which is dedicated
to doing data cleansing. So what I’m gonna do is
actually just take this URL, the one I just pasted here,
go into Power BI, connect to web. And I’m just gonna click Connect. This can ask me to enter a web URL. I’m just gonna paste it in here and
I’m gonna click OK. What Power BI is gonna
do behind the scenes for me is actually connect to
the website, download the data and say hey, here you go. Here’s the data for you. And now I’m gonna perform
some cleaning on it, it’s gonna come in XML. I wanna actually get in the right
format that I actually wanted. And what happened here. Please wait while we
establish a connection. Hopefully the Wi-Fi will improve. And as you can see, the data is
not really meaningful for me. It’s only got one record. It’s only picked up one
record by cert detection. So what I’m gonna do is
I’m gonna click Edit and actually go ahead and
make that data more useful. So one of the things, a great thing in Power BI is I can
just click on columns I don’t need. So I don’t need these
first three columns. I could say I might wanna
keep the updated time but I’m just gonna delete them. What I’m gonna do is click
this little icon here, this on the entry tab which says
hey I’ve detected there is a table underneath this column,
do you wanna expand it out? Which is effectively
what it’s saying. So I’m just gonna click here and
just say expand. And it’s gonna say these are the
columns that have been expanded, do you wanna select
specific columns? I’m just gonna click OK. I’m gonna have to do this a couple
of times, so bear with me. I’m gonna click the content tab and
expand arrow again. And then go to the content tab and
expand arrow again. This is just because someone
made the XML highly nested. I have to keep expanding out until
Power BI gets me what I want and then I can expand it again. And then I’m gonna
expand it one more time. And now I’ve got the data
that’s in the underlying XML. I’ve got the country, I’ve got the
medals, I’ve got the frequency, and I’ve got the date. The basic information for
the Olympic dates I need. So I’m just gonna click OK and
expand them out. And as you can see, I’ve got
country or location, I’ve got when the Olympic games actually
will held, summer or winter. I believe that says total,
whether what type of medal was, total aggregation medal or gold. And then, I’ve also got some
other information like the year of the Olympics were held and
the count. Some arbitrary value depending on
whether I’m looking at medals or whether I’m looking at ranking. So what I’m gonna do here is
something that you can easily do in Power BI which if you use R or a Python sometimes doing this
can be somewhat tedious. I’m just gonna select some columns
and I’m actually gonna click delete. Very easy and as you can
see at the right hand side, I’ve got the steps of cleansing
of before the Power BI dataset. This will be always repeated. So if I said just turn refresh,
when the next Olympics comes out, I don’t have to change my Power BI
report, it’ll automatically update. Just by looking at that web URL, that web URL changes
with the new data. These sensor steps are just done
now, will be maintained and done in every refresh
of that dataset. I’m gonna rename this column
because sadly with XML, it does some really weird things. So I’m just gonna rename this. Let me find the real name for it. I’m just gonna put here country. I’m gonna put type of game for
this column. And this is somewhat tedious
sometimes, I apologize. And I’m gonna put here
this is the Type of medal. Then I’m not too interested in this,
so I’m just gonna wipe that out. I’m not too interested in this,
I’m gonna wipe that out. Now I’ve got a date/time field which
has been picked up as a string. The great thing in Power BI is
this thing could transform. And I’m gonna transform
into a date/time. And it’s automatically
picked up a date/time field. And now what I’m gonna do is,
I only am interested in year, I’m not really interested in time
of year for the Olympics, so I’m just gonna convert
this into year. Very easy, and
I did that in a few simple clicks. If I was doing this in R or Python,
I would have to somewhat rememorize what the actual command and let me
just put here year and continue on. The final columns is value. This depends on what I’m
actually looking at filters. I’m just just gonna put
here a value for now. And then the other columns, I’m just gonna wipe out just
because I’ve no interest in them. And now what I’ve done is I’ve gone
from something highly nested to an XML that’ll probably take
me a couple hours to do in a different language. In Power BI I’ve done them
in couple of minutes. And in Power BI all you need to
do now is to say close and apply. Basically take that dataset,
now let me do some analysis on it. And this will just take
a couple of seconds to repeat the steps on the full XML. And now it’s done. A couple of things in PowerBI you
might wanna realize is if you guys came from an SSS background, you’ll
have to normally set the data types, PowerBI will only pick them up for
you but sometimes it doesn’t. So I’m just gonna click on value, go
to modelling tab and just say, hey, this is a number not a text. And what I do here is I’m gonna say,
I think it’s a whole number, I don’t think there’s
any decimal places. It’s letting me know that there are
some impacting changes that’ll need decimal places there. And there you go. What I’m gonna do now is I’m gonna
say hey, I wanna look at the year, and I want to look at the value. For now,
this may not be that meaningful, let me put year on the axes. Try that again, there you go. This is not meaningful because
I’m actually looking at a of everything including gold, silver,
world ranking results put together. So let me just put some
filters up very quickly. So I’m just gonna put country here,
and turn this into a slicer and make this into a list,
I’ll do that in a second. Let me just wither out the way. I’m gonna put the type
of medal here as well. Convert that into a, I’m gonna
convert it to a drop down, actually. That might be easier to see. And just gonna put that there, and I’m gonna just nicely
format this thing. I could spend hours here doing
this but I’m not going to. And I’m gonna put the type of games. I think I should do that as well,
yeah. And let me just
quickly minimize this. And the great thing about Power BI
is I can just quickly drag and drop and really get the results
I’m really looking for. So I’m just gonna condense
this dropdown as well. And so this time what I’m gonna say
is I’m only interested in the summer Olympic games. I’m only interested in their world
rankings, so Olympic ranking. And I’m originally from the UK,
as you can sense from the accent. I’m actually gonna look at how Great
Britain did over the last couple of years. And if I select great Britain,
I can actually see how their ranking has improved
over the last couple of years. I could start comparing
this with other countries, I could look at their gold
medal counts as well. So actually, try looking at Olympic
ranking let’s see how Great Britain has done in total gold medals. And as you can see the program that
is implemented in Great Britain is actually helping improve
our gold medal count. And again I could go
really savvy here and I can start comparing different
countries, draw more graphs and actually do a deeper
level of information. The great thing is I was
actually gonna do this but I think you guys hopefully get
enough about Power BI that I was actually gonna mix it up
with the IMDB movie data. And see if there’s any correlation
between the year the movie comes out versus how well we do on our meta
data but I won’t do that today. And so
once you’re done with the analysis, I’m just gonna save this file. Let me just save it somewhere nice. I’m gonna put here that
this is the Olympic demo. And I’m gonna save it here. And once you save it,
the great thing is, in Power BI, you can actually
directly publish it to the service. So the desktop tool and the service
are directly linked to one another. And so I can now publish this,
share this out with my organization. So let me just show you quickly
how you’d accomplish that. So I’m just gonna say publish. And it says can you sign in, so
I’m just gonna try to sign in. Hopefully it won’t ask me to
do two factor authentication. Wrong e-mail, And
I’m gonna click Sign in. And so it’ll go ahead and
sign me inside my organization. Hopefully, this will go away,
this white box. And now,
allow me to sign in to my Power BI. And if you guys use Office 365, work spaces where you can
actually group things together? So I always create myself a test for
this demos. So I’m just gonna find
test more group and just stick it in here for now. And now this will go ahead
save the Power BI file now, uploaded to the service
where I can now share it, set up daily refreshes of this
dataset if it was able to change. And I can also do some other
interesting things with it. I can also see quick insights,
but I’m not gonna do that today. So I’m just gonna go
into, and let’s see if the report has
actually shown up there. So if you go on,
you click Sign In in the top-right corner, You can just quickly go
into the Power BI Service, and it’ll look very familiar
to Power BI Desktop. It doesn’t have the same number or level of capabilities as Power BI
Desktop in terms of cleansing data, but it allows you to see and
analyze the data. And so you can see a lot of people
share dashboards with me every day. This is the UAT environment for
Office 365, if you are very interested. Let me just find the workgroup
that I was very interested in, so I’m just gonna expand my workgroups. Click on more cuz I have quite a lot
and I’m gonna go to test more group. And hopefully, you’ll see the report has shown up
there the Olympic demo’s there. So now I’ve got my report here. Got my thing in. What I can do is now
make a dashboard, which is basically a collection
of different reports tiles. And if I show you an existing
dashboard, I had an MS band demo. I can just create
different tiles and just create my unique experience
of what I want it to look like. This is quite like Pinterest where
you can pin different images onto your main favorites screen and
analyze. And just every morning when you
come in to work, you can just see the high level information that
you’re really concerned with. And then if you notice an anomaly, you can actually drill
down into reports. Very similar in here,
I can also set up daily refreshes. So I have this notion
of schedule refresh. So I can say, hey,
schedule everyday, every week, every month, every year. I can also analyze in Excel. So if I’m very Excel savvy or
people in my organization are. If I share this with my business
user and my business user notice some mistakes sometimes they wanna
actually do the analysis themselves, they don’t wanna wait for me or sometimes they are on their
weekend and they actually like. I wanna do some other insights,
within Power BI Service, I can just click this button they
will get the same data that I always is dealing with and
then they can do their own analysis. And this is the third way for everyone in the organization
is entitled to be BI. I can also do quick insights but on this dataset it might not
come out with that real results. And what quick insights does
is it actually goes ahead and finds you nice correlations
in your graphs. So it actually says hey,
did you know this? And actually comes up with some
statistics about your data. It depends on the type of dataset
that throw it, on the Olympic dataset it will come up with
just telling you who the first third persons are in
terms of what ranking. And, it will be quite easy to
find this information out. What I can do, as well,
if I go to the Olympic demo again. I’m just gonna pin this
graph onto my dashboard. I’m just gonna say, demo dashboard. I can actually use Cortana. Even if I had Cortana on my phone,
which, if I publish this dashboard, I can access Cortana from my phone. Or I can just use Power BI and
natural language query inside here. So let me try this out and
see if this works today. So I can actually ask questions
about my data, and I can say, hey, I’m interested in where
Great Britain and, well, actually let me ask
a different question. Who’s world ranking number one in 1996? Let’s see if it decides
to answer this. I spelled world ranking wrong so
it may not help there. And let me just do world ranking in
order and let’s see if that helps. If it doesn’t,
maybe I’ve set up my model wrong. Okay, so it’s decided to
return me all the data. Maybe it wasn’t called world
ranking, maybe I didn’t that column. I will give up here. What you can do is actually ask
natural language query, if you train it and give it the data set, I think
I named the column incorrectly in the original data set, which is why
it’s not able to answer my question. And I can ask the questions
through my PC device as well. So that’s Power BI the service and Power BI Desktop working
harmoniously together. The other great capability of Power
BI is I can set up security, so I can make sure that only the right
people in my organization see my data. I can also set up
an on-premise gateway. So if my data sits on-premise and I
don’t really feel comfortable moving it to the cloud, like a lot of the
financial services, you can still use Power BI as a web front-end for
your reports and dashboards. But you can use the on-prem
data source and never bring the data source
directly into Power BI. And so that way you can get around
some of the security restrictions that are maybe set inside
your organization. And so
that’s Power BI in a nutshell. What I’m gonna do is quickly go back
to my slides and try to continue. Does anyone have any
questions about Power BI? It’s quite hard to see your hands,
just shout if you do. And generally, Power BI’s constantly
adding the service updates once every week, the desktop
updates once every month. And we have a lot more capability
there, I’m gonna show you later on. We have something
called custom visuals, where actually, if you’re good at
D3JS, you can actually define and create your own custom visuals. If you’re good with R,
you could also create R visuals, so you can see there is a nice scatter
plot of R on this diagram as well. And generally,
you can make whatever you want. Power BI is a platform for you guys to go in and mold what you
want in terms of your own analysis. And that’s what we’ve kind of done
with the promise of Power BI. If I move ahead, the five promises
for Power BI is really we are trying to engage our partners to actually
drive a better pilot community. And by doing so, we’re allowing customers to quickly
find partner-based solutions. And I’ll show you that
a bit later on as well. Obviously, it’s to help making
you more data cultured, making you drive better
data-driven decisions. And I’m gonna show you more
of the right side later on, around solution templates and
partner programs. So that’s Power BI. If you guys used SQL Server
Reporting Services about four or five years ago, you may have
noticed that not much has changed. In the latest version of
SQL Server Reporting Services, we are glad to announce
that we have changed and overhauled the complete UI for
SQL Server Reporting Services. So that’s to make it
more mobile friendly, to make it more
aesthetically pleasing. And what I’m gonna show you today
is actually a demo of the new SSRS preview that I think
released last week. And so, SSRS is going for
a long learning curve. SSRS eventually will allow you to
embed Power BI reports in SSRS. So for people who don’t even wanna
use Power BI service online, you can actually embed Power BI reports that
I’ve just authored on my desktop tool and actually upload them into
SSRS, which is completely on-prem. So I don’t have to even ever go
into the cloud if my customer’s not ready. Or if I’m dealing with governments
or if I’m dealing with organizations just very locked down in
their thinking in cloud. And so we don’t wanna
alienate those customers. We’ve made SQL Server Reporting
Services, made huge bets on this. And what I’m gonna show you is
actually how you can spin up the demo yourself, the one
that released a week ago, and how you can quickly get there. A quick show of hands, how many of
you actually have Azure accounts? Is it quite a few of you? So if you sign up for
an Azure free-trial account, I’m just gonna show
you very quickly. If you go to, one of the great things we do
is we keep this very up-to-date. And we actually release VM images
of SQL Server Reporting Services. So if you ever wanna demo this to
your customer, if you wanna demo this within your organization,
you can actually go in and click on Resource Groups, which are like
collections of resources in Azure. I’m just gonna click Add and just show you how you’d
go about doing this. I’m just gonna add a new item,
not that. Click Add here. And what I’m gonna type in here is
SQL Server Reporting Services in my most recent. And if I click this, what it’ll
say is hey, I’m gonna create you a SQL Services Reporting VM on your
subscription, which will allow you to carry out the same demo
I’m going to show you today. And so when you click Create,
it’ll show you the Azure option. If you’re very familiar with Azure,
say, can you provide me a VM name. A username and password to login to
your SQL Server Reporting Services and some other basic information. I’m not gonna go for this option cuz it can take
about ten minutes to spin up. Instead, what I did before the demo
was actually spin one up for you guys and let me just quickly
see if I can navigate to it. So I saved the link here. When you set the VM, all you need
to do is navigate to the VM. And just go to the reports page,
so just this URL here. It’s gonna go to another
browser page and just go here. And it’s gonna ask me for my
username, password that I just set. So I’m just gonna click Log In. And so SQL Server Reporting Services
has come a very long way. If you ever remember the old
version, you’ll notice that SQL Server Reporting Services
drastically changed and will continuously evolving
over the next couple of years. Acquisition of data then is fed a long into
SQL Server Reporting Services. So as you can see, a lot of that
stuff that we’ve acquired from data analysis actually come into
the SQL Reporting Services portal. So you can see that I can now
have KPIs, so I can see hey, how is North America doing if
I’m the North America region. This demo is actually set for fraud detection so I can quickly
find out if there’s any fraud happening within my organization
in terms of if I’m managing sales. So I can look at my
corporate scorecard, and again I can just look at pages,
quickly analyze the information, quickly see what’s happening
within my organization and it’s very interactive,
very similar to Power BI. It doesn’t have the same
capabilities of Power BI and that’s why we’ve integrated Power BI
directly in SSRS, which I’ll show. And yeah, I would say the amount
of investment that’s gone into since 2008 SSRS,
it’s come a very long way. The Paginated Reports
are still there, so if you’ve got queries that
are running paginated reports, these will still render with a new
fresh look and feel of modern BI. And I can again do the same
capabilities I had before. I can print the reports out. I can zoom in. I can interact with these reports
very similar to what I would do before. This query happens
to render one page. If it didn’t,
I can click the next cycle and I’d see different pages
of the same report. I can also pin tiles
to my Favorites. So I’ve pinned some of my key KPIs. In this case, I was only interested
in one KPI, the same way that you would do in Power BI, the same way
that you would treat a dashboard. I have a Favorites page, where I can really suit to my
own needs and requirements. The one thing that we’re really investing in is actually making
Power BI a second class citizen for SSRS, or sorry,
a first class citizen. And now if I click on some of
the reports that come with the demo, I can now see Power BI
embedded directly inside here. The data is on-premise,
Power BI is on-premise, I author reports is on-premise. And I can also connect to
outside data sources and bring them in using
this tool as well. And so
this is part of the first demo. Again, I don’t need
to convince you guys, I can interact with it exactly
the same way I do with Power BI. It’s a like-for-like match. A lot of investment has
gone into doing the effort to get this integrated into SSRS. It’s taken some time, and so this
is why it’s still in preview and we’ll be hopefully seeing
this feature very soon. And so SSRS will become the hub or
the Power BI service for your on-premise. And Power BI will act as the glue
between hybrid data storage. So you can have your data sitting
on-premise, in the cloud, and you can use the gateway. And then, link everything to get
actually one view of all your data points. So, that’s SSRS. I can show more later on if you
guys are really interested in SSRS. I would highly recommend you to go
ahead and spin up the VM yourself. And then once you’ve done the demo,
you can just tune it down. It shouldn’t cost you that
much to run that example. And if you’re using the free trial, it’ll actually be included
in your free trial.>>Which version will
that be released in?>>So, SQL Server 2016 came out. SSRS I think will come out
under SQL Server Service Pack. It’ll come as a separate download. At the same time when SQL Server,
it’ll be 2017. We haven’t committed to a date
because we don’t know when the date will be for GA. And that’s something that if you
wanna follow up, I can help you get in touch with you the PM who’s
managing the release cycle for that. Any other questions for SSRS?>>None of these features will be
available in previous versions of SQL Server, will they? [INAUDIBLE]
>>In SSRS you mean?>>Yes.>>So the SSRS preview, you’ll be able to install it on top
of SQL Server 2014, if that’s what. You’ll be able to install it on top. It’ll be like a feature pack on top. But it will be a separate install. And the reason why is I think it
will support up to 2012, actually.>>But you’ll still get
all these new features?>>You still get all
these new features, yeah. You’ll be not limited
in those new features. There was a question in the back.>>So
is this the I’m sorry confused, but is this the SaaS model
offering of the SQL Server?>>No, so when you publish the VM, it actually gives you
a public endpoint. The VM has port 80 open, so I just go into the VM and I go into
port 80 and it says Go to Reports. And if I was on-premise and I was
to log this way behind a firewall, only people inside my organization,
inside my intranet could see this. I made it publicly facing just so
I could do this demo nice and easy without having to
remote in to a VM, and then showing you guys that
I’m actually in a VM. It’s purely for my convenience. Yeah, I’m sorry for the confusion.>>[INAUDIBLE]>>So when you say customization, which specific customizations,
have you an example?>>I actually write my
own business logic by>>Yes, so SSRS, the 2016 and above has a new format. I think they changed the format name
just to indicate that it’s 2016. The way you generate reports is exactly the same way as
you would generate for 2014. You have new controls, new buttons, new ways to
render reports, new visuals. But they way you would
normally do that, the way you can embed
business logic in it, the way I still page new
reports was still there. To update your old, old reports,
I’m not sure if there’s gonna be a utility to do that, but
you can still do your own reports, do the same thing that you
did in the 2016 as well. So SSRS hasn’t changed, what we’ve
done is effectively introduced more capabilities, because SSRS went
through a period where it didn’t really get much updates,
if I’m really honest with you. And now we’ve put a lot
of investment inside it. Is there any other questions for
SSRS? Okay, so that’s SSRS. Now let me go back to the slides. A lot of this is very demo heavy,
my slides, just get two or three slides. So yeah I went into the SSRS demo,
let me just move away past this SSRS demo, and
let me now talk about Scale Out. One of the things that people
tend to get really confused with when dealing with BI is once they
fall of the cliff they never return. And that is they set up
some nice looking reports, nice looking dashboards, but
as the data sizes grow, grow and grow eventually their
reports are rendered useless. And then they have to spend
weeks to change the format, change the data scheme
because they did not scale. And with Power BI service, there’s a one gig data size limit on
if you just use a service directly. And we acknowledge that limit and
said hey, we can help you overcome that limit by you using
SQL Server Analysis Services. Last year we announced
that we were gonna make SQL Server Analysis Services
a pass offering inside Azure. And what I mean by that is I can
spin up SQL Server Analysis Services without requiring a VM. I could spin it up, pause it when I
need to, when I wanna minimize my costs and only run it when
my business needs it. And so that way I can keep a tight
hold on my costs in my organization. I don’t have to worry about patches,
updates. This updates for itself because it’s part of
the pass offering inside Azure. And I can keep an eye on my costs
which is the most important thing. And a SQL Server now is set up to
allow you to deal with 100 millions even billions of records at a time. And so SQL Server now is basically
what you find in SQL Server 2016, it’s the same technology. We’ve just taken it out,
put it inside Azure. Provided a neat
pricing around it and effectively allowed you to quickly
spin up inside the organization. So that’s one scale-up model. If you notice that the refreshes
are taking too long, the data sizes are just too large. You can use SQL Server and Azure Services to create yourself
a tabular model inside SQL, like you would do in
SQL inside Azure. And then access it the exactly same
way you would do, and I’m gonna show you how you can spin order
inside the Azure portal in a second. The next section is, how many of you have heard of
the definition called a data link? Okay, so
I’ll just briefly go through it. So, a data link is a place where you
basically dump terabytes of data in. You dump in semi-structured data,
structured data, any type of data, but the point is,
is that you just dump data in. And then, when you wanna answer
the question that you need to ask, you then do
a schema-on-read operation. It’s the same idea as Hadoop. Hadoop has the same principle. You dump things in a file store. The file store is
usually quite cheap, and then when you wanna do the
announcements on top, you pay for the computer to do the job
that you need to run. The problem with Hadoop, of course, is you’d end up having
to buy incredibly higher long line of infrastructure
that requires it. With Azure Data Link, we’ve actually
don’t require you to do that. You just write your query that
you wanna do on top of your data. You submit that to us and
you only pay for the compute for running that job. So if I was running
a job that required, like I ran one this morning. It ran through about 50 meg of data. I charged myself, I think 0.0003 cents of a dollar,
just to run that job. If I was dealing with terabytes of
data which we do internally for the Bing search logs. We use Azure Data Link, and
to really go through that we only charge ourselves about ten dollars
to really go through that. And it depends on how many
states you’re dealing with, but it’s meant to be a scalable model. So today I’m actually going to
spend some time actually go through Azure Data Link because I think it’s
very important in the BI world. As you deal with more and more data requirements,
like adding in telemetry, adding in the logging, especially
if you’re working in the IoT space. The data volumes
are just too large for you to just start piping into SQL
because you didn’t have to decide in your schema when you put it in SQL. Sometimes when you wanna
answer a question, you don’t know what your schemas
will be until you actually figure out that this is
the question I need to answer. And so with the changing
roles of business, Data Lake will become more of a pivotal
role inside of a organization. And once you get the data out
of Data Lake, you do your query, you can then pipe into SSS and then still provide your business
users with a fluid PowerBI, reports, dashboards, the whole lot
as you normally would. And so I’m just gonna spend some
time actually going through SQL Server Analysis Services first. Just really show you what it
means in the Azure portal. And then go through Data Lake and
how you’d submit your first job. So let’s start with SSAS. So as I mentioned it’s the same
technology that you would find. SQL Server Analysis Services
in Azure is only using tabular mode at the moment. If you want to vote for the idea
to support multidimensional, just go ahead and use a voice. We highly listen to
our user feedback and please vote for your idea. It uses the same processing engine
DAX, if you’ve heard of DAX, which is basically a syntactical
layer that allows you to write your own measures,
your own calculations. So if I want to calculate
the year on year growth for my organization, I could write
a simple DAX expression to do that. It’s all in memory. It can serve lots of my users. It can work with both on-premise and
hybrid scenarios. So if my dataset is in cloud or
if my dataset is in hybrid, SQL Server analysis services for
Azure has a gateway story as well so I can connect to
wherever my dataset is. So I’m not really bound
by where my data is. I suppose that is it
from Analysis Services. I’m gonna go in the Azure
portal when I go through the Azure Data Lake in a second. Azure Data Lake,
as I’ve kind of gone through, is this notion of a store and
you have the analytics there. They work in harmonies
with each other. You store your data, you pipe
your data, telemetry data, in. This could be in terabytes,
even petabytes of data. There’s no restriction
in Azure Data Lake. And then when I actually
wanna do the analysis on top, I can use any utility I want. I can use something called U-SQL,
which just a SQL-like syntax, and I’m gonna go through that today. Or I could use Spark or any of
the other open source platforms. But the key thing here is I’m
only paying for the compute. I’m never gonna go in and say a 12
cluster node with one head node or two head nodes for
high availability. I do not need to worry about that. I can now finally forget that. And from the Azure Data Lake store
I can really do a range of things. Once I get and do the queries
inside Azure Data Lake, I can start doing things like
machine learning, real-time insights, I can do interactive
queries and I can do batch queries. So if I really wanna really
get the answers really fast, I can run the instructor mode. If I want to do batch queries I
can run it in batch mode as well, and that’s what I’m
gonna show you today. And so typically you wouldn’t
find BI people talking much about Azure Data Lake. You’ll only hear the big data people
talking about that data platform, specialists talking about that,
but I think it’s very important. And so let me just quickly come
out of this slide deck and let me go back in the Azure portal,
my trusted Azure portal. And let me go back to
my resource group. So first MIT spun up these
resources in Azure today. I won’t bore you. It just asks for my basic
information and my password. And what I’m gonna do is I’m gonna
click on AzureAnalysisServices. So it appears under my Azure
portal like it normally would. And there it is. I have a Portals button which when I
click Portals I will no longer get charged for
until I click on Portals and run. So this allows me to optimize,
I can create a job which basically says every weekend,
turn off my announcer services. And just from Monday to Friday,
only turn it on. This is something you couldn’t
do with AnalysisServices before, you normally have to buy the
infrastructure, and there is no such thing as pausing the infrastructure
in an on-prem environment.>>[INAUDIBLE]
>>So in AnalysisServices, the pricing model includes
the tablet model that sits on it. But you need to give it the data
where it comes from and that could be SQL Azure so
you’d pay for SQL Azure separately. It could be on premise and
you’d pay for on premise separately. AnalysisServices kind of
sits as a separate entity. So as you notice, the only
information I really need is the SQL Server name, so
AnalysisServices name. And I’m just gonna copy this,
and I’m gonna open up SSMS. I’ve already connected to it. Let me just disconnect to it and
I’ll just show you. I’m doing no different. Analysis services,
would be very hard for you guys to really notice the difference between
on prem or Cloud AnalysisServices. They use the exact
underlying technology, and the tools treat them no different. There’s no such thing as saying, hey, I can do this capability here
but I really can’t do this here. We tried to make them, effectively, feature-powered here
as much as we could. And so all I need to do is type
in my server address here. When I click Connect, it’s gonna
pop up with the OF screen, which is my Windows credentials. It actually uses
My credentials. Because I already signed in before,
I’d go in. And I can then view my Databases. I can view my management. And I can assign permissions
like I normally would. I can go into
SQL Server Data Pools and you’ll see exactly
the same way as well. Make my own SQL Server now for
the purpose of today I’m only actually gonna create a model for
you. I didn’t think that was much
useful for you guys’ time but I would highly recommend you to just
go and spin it up in Azure, go and create your first database and just
see the performance for yourself. Like you can set up your own
processing schedules and optimize it for
really your own organization. So that’s Azure analysis services. It’s a great feature. A lot of customers highly
requested it and I think it will directly help and help provide
that scale out story for Power BI. Now moving aside from
analysis services, I’m actually going to go
into Azure Data Lake. And with Data Lake,
Data Lake comes with it’s own store which is really cheap in
terms of it’s data storage. and in concert with the data
analytics analytics engine. And so the data lick store,
I’m just gonna click on it, I call it tech summit two. Tech summit was already taken. Someone might have taken it,
and what I’m gonna do here is, I’m just gonna go
into data explorer. So, the great thing is that I can
now put some sample data in so I put some sample data in here. Samples and I’m gonna go to data. I’m gonna move to ambulance data
the time the demo was looking at the data and I’ve got lots
of range vehicle data here. As you can see I’ve got a lot
obviously I have just drawn this for the demos so that you guys
weren’t waiting an hour for me to analyse information but
of course, I got a bunch of data about
different vehicles here. And what I want to do is actually
combine all this data real easily and write a SQL query on top, maybe
do a course, maybe do a group by and aggregate that data so I can do
some analysis and tabulate it. And so in data leak, the great thing is what we’ve done
is we’ve made the portal be kind of a first place for you to go into and
try to analyze this data. So what I’m gonna do is, I’m just gonna close this,
gonna close this again. And I’m gonna go to the data
store analytics engine. So just click it and the first
thing I’m presented with is, hey do you wanna manage the user
permissions of who can submit jobs? You can see your job history and what you can see is
some sample scripts. And you can also save fewer jobs and
actually create a new job. Now I’m not gonna memorize
a syntax that I wrote just five minutes ago
before I came onstage. So I’m just gonna go
to a previous job and actually just click
duplicate script and I’ll walk you through the script. And you guys can tell me, this is
very much like SQL command syntax. So let me just walk
you through this, if I can zoom in then maybe
that might help you guys. So there’s some slight nuances,
differences between SQL and U-SQL, mainly because U-SQL relies on more
of the data being unstructured, semistructured, so we have to
introduce new capabilities to handle those kind of capabilities. And then that’s where there’s
some nuances in syntax. And so what I am doing here
is that I’m saying @searchlog which is a variable,
searchlog equals extract and the data in my data store
could be unstructured. In this case, I am saying,
I’m reading a .csv file and the .csv file is basically saying
take anything that has the vehicle file path. Star anything after it until
which the date time, and .csv. So I wanna it’s consume anything
that’s csv in sample data. I cannot support another
world client and say in any sub root folders off,
find any further matrices. So I could take potentially all the
data for all the ambulance across the USA across the last ten years
if I really wanted to, but for the demo I didn’t wanna
sit you there and wait for you guys to watch a job to complete. The data in this case, in the CSVs
were dumped in a specific format. I’ve now just decided to provide
a format along the data. I’ve got vehicle ID, I’ve line number when it
came into my log file. I’ve got the time, the longitude, latitude of the actual vehicle
when it was responded [INAUDIBLE]. In this case I’ve got an IT
device in every single ambulance across the US. I can monitor where that
vehicle is at any time. I’ve got the direction in terms
of whether it’s travelling NorthEast North. Ive got some values that I
might be interested in but I’m not gonna spend too much time
let’s [INAUDIBLE] go ahead and actually drill into
the sample dataset. So if I go to the Data Explorer and
let me see if I can access the sample data and just
show you what the data looks like. So I’ve got ambulance data, I’ve
got vehicle1, vehicle2, vehicle3, vehicle4. Let me just go into
one of the CSV files. You’ll see it’s very simple, very typical scenario
that you might have. You’ll have log files,
telemetry, coming in. In this kind of format, and now you wanna really
analyze this information. If I go back, and
let me just zoom out this time. Hopefully this won’t affect
my browser too much, and let me go to the job. All I have to do now
is click Submit Job, it actually tells me how
much this is gonna cost me. This is gonna cost me $0.03 a minute
to run because the infrastructure I chose was a standard infrastructure,
and you can decide if you wanna run this across more computers and
make this job finish faster. And you have a little slider
to choose a scale model. I just chose the most simplest basic
one just to show you the example works even in the most
basic scenario, and I’m just gonna click Submit Job. And what this will do now is tell
me how far my job is going to do. It actually gives you a nice
pictorial example of where it’s gonna get the data from
the different data sources. In this case,
it’s only gonna show me one node, which is it’s reading it
from one location, and it gives me a nice little timeline
of how the job is coming along. As you can see, it says hey, I’m going to do this extract step
and then I’m gonna do an aggregate, very similar to the way
the Hadoop map produced works. The first stage is a map. The second stage is a reduce,
if you guys are really purists. And then it’s gonna
output vehicle output, which is all the data,
shifted together in one view. I cleared it down to where I’m
looking at specific vehicles or specific time periods. If I’m looking at anomalies for
my organization. And so this is something that you
couldn’t do, like imagine if I information, I would not dream
of putting that insight SQL. And unless I had specific questions
were also put in SQL if I had a daily report that
would be coming out. In this case I know specific day. I actually wanna go ahead and
analyze it. And it’s a great example of the
scenario if you’ve ever heard of it The guys that fit,
I think they in California. And so
they use everyone’s fit that data to figure out where the center or
the earthquake was. And they did that just by going
through based, the fact that they stored over data in this role format
help them quickly find insights. And they will first write
the article blog post out saying we found out where the centers
by working out which Fitbit devices were being triggered off at
what time versus what happened to the others, and they were there to
pinpoint the accuracy using Fit. And this is some of the things
that you would never store in SQL. You never say, hey I’m gonna use
this to predict earthquakes. You just wouldn’t do that. And so the job is finished. The great thing here is
I can just go in and just look at the sample output. So I can just quickly analyse it. It I wanna debug my job as
well you can use Visual Studio we have tools available as well I’m
using the pool just to show you how quickly is to do. But I can actually
go in the pool and just see how the sample
app are doing. And then I can just pipe this data
into SQL I can pipe into SQL server and Azure services. I can actually get
Power BI to connect to data link directly as well. And so it’s up to your
organization’s needs and requirements but Datalink is
there to allow to analyze the further end of
the spectrum big data when you’re really talking about
the terabytes and petabytes of data. And in Microsoft we use this to
analyze our search log data. And the search of data is just huge,
and for google as well is very huge as well. You’re dealing with data that you,
you just wouldn’t store. You’re not analyzing in daily basis, you’re analyzing it probably
in a month by month basis. But you still wanna keep
the historical data, that just incase situation arises. Maybe there’s an anomaly
in your search results, so you wanna go and optimize. And for machine learning as well for people who are very tech
savvy data scientists. Having all the historical data is
a dream come true for you guys. And that’s a very important
thing for these MO guys. So that’s Azure announces date
like any questions about these two before I move on? Yeah.>>General question>>[INAUDIBLE]>>[INAUDIBLE]>>So there is definitely guidance. I think the discrepancy here is
when you say semantic model, it depends what you
mean by semantic model. Data Lake is used for answering. In this case,
this was a batch query. This is for me to answer a question
>>Or deal with. What I would normally do with
is get the data in a form that I would normally pipe into SQL,
into a schema in SQL, and then I would use to pipe it to
basically connect to the SQL to allow multiple users to fire
different types of queries. The great thing about SAS is
because it’s a cube-like structure, it’s a column store index
If I’m trying to do or filters like I was doing with
the Olympic dataset, to go to and get the result within a second, SSH
was designed to handle that specific requirement, where some user
may come in and say hey, I want to understand how doing,
how Africa as a nation is doing. And so those kind of questions
are not normally preempted, but they’re in the data model in
the way I can just quickly provide the slice and dice to that cube. And that’s the way you’d use SSS. SSS is kinda the final point before
the business user connects to you. They’ll fire the queries to the SSS. Whether you should use SQL or
DataLink for your storage mechanism depends
on the need of your business. And normally, your store SQL
was only have a backup job. You’ll have a backup job that
just say hey, every two months, can you date the and
put it in the data warehouse? And then when you come to data
warehouse data link group, it gets kinda mushy when you should
use which of us is the other. I suppose if you’re
using data warehouse, you’re normally quite comfortable
with the data warehouse tools if using PDW or SQL Data Warehouse,
Azure SQL Data Warehouse. You’re normally quite comfortable. It mainly has to do with the skills
sets within the organization. When you use Data Lake, and
Data Lalke’s recently been a new phenomenon, this initial Hadoop is
when you’re dealing with I’ll say more of a batch, when you don’t
know the question you’re answering. You want to just
analyze lots of data, it’s not in the format you need it,
you want to do lots of cleansing. And so, you’re kind of
bringing ETL in one cycle. In SQL, your data is
quite heavily organized. You’ve already got the tables, you’ve already got
your customer table, you’ve got your customer location,
and you’ve already got the tables, and you’re already working
within that constraint. And so there is definitely guidance,
but I say it depends on the specific niches of what you’re comfortable
with, in terms of an organization. And it also depends on what
you guys are looking to do, in terms of answer. If you need this question answered
always, you probably wanna do an ETL process, have a Data Lake job that
kicks off, and then writes into SQL. And then you would use SQL and
SSS together. I kind of see them as
harmonious with each other. But it’s a good question. One of the great problems of
Data Lake, and generally, the technology in the last
ten years is there’s so many startups have made so
many different tools. And sometimes, it can be a pain when you’re talking
to your customers saying you might use this, you might use this,
you might use this. There’s about five
different streaming technologies that I know of. And sometimes, the customer can get
just overwhelmed, and sometimes, you just have to say look,
we’re gonna go with this, unless we notice we
hit the limitations. And I used to work in consortium for
three years. Sometimes, a customer
would be concerned about the technology they would use. Sometimes, they just don’t
wanna go down to lockdown. They don’t wanna be locked down for
specific infrastructure, specific software. I said that’s the great
thing about using cloud. If they want to go away from,
Data Lake is a different example. Data Lake allows you to use Spark,
Hadoop, or any other proprietary
tool you’ll use. Data Lake is just kind of like
an intercessor layer saying, hey, I will allow you to use
the same cost model, but you can use whatever you want. And so, I see the need for, and they are different data guidelines,
so when you should use which. But I’d say one of
those things that, with experience will come with which
one you’d use in which situations. Sorry, the answer’s
not that helpful, I’ve used very few, I’ve used quite
a lot of Hadoop implementations. Azure Data Lake was still in preview
at the time that I was using it with customers, so customers didn’t like
me using preview products until just mainly for a support cycle. And so, depending on the needs
of the customer depends on, which we’ll use. Any other questions for Data Lake or
Azure Analysis Services? Okay, so, I’m gonna go back to my
slides, and I’m gonna do something a little bit more fun now, which
is the advanced analytics section. So, people,
when they say business intelligence, they normally mean answering
the question of what. What happened in my organization? So, let me look at the data that I
have now, and let me go figure out what happened, so I can go and
tell my business manager, or my higher ups what happened
in this situation. Why did my sales hit this problem
is a different question altogether? And so, this is where you talk
about this advance analytics, answering the what,
the how, and the why. And so, in Microsoft, we have kind of provided
two kind of models for you. One is the take an API
that we made for you, a pre-built model that we’ve got
our data scientists to build. And then you can just plug and
play them into your Power BI report. And I’m gonna spend a lot of time on
this with a demo of Power BI showing this very capability. We’ve also shown that sometimes, you have got model highly
specialized for a customer. You wanna just go and build it using
your data science team to go and build a highly specialized model. And that’s why we
have something called Azure Machine Learning Studio. Now, if I have time at the end,
I will go through this. But I’m not gonna spend too much time showing you how you can
make machine learning models today. I think there’s a separate track, separate session that I
should go through that. And so, when people normally
refer to business intelligence, I normally see them asking the what,
but never the how and the why? And that’s where advanced analytics
really comes in, really shines. It’s saying hey, I can understand
what happened to your sales, I can understand how it happened,
because of this specific, the weather conditions outside. Now, wouldn’t it be great if I
could understand what contributors actually factoring to that cause? And If I can predict that,
could I then better estimate and handle my resources
in my organization? So, let me give you an example
of a customer that I work with. Let’s take a typical supermarket,
if I could predict my sales for my supermarket for each product,
I could then better optimize my storage requirements of
what products I order when. And you think it sounds like
a very trivial problem, but it’s actually a very expensive
problem for supermarkets. Storing an extra pair of produce,
having excess storage inside my supermarket is actually
a very costly operation. They spent millions on this, and
just making a machine lending that has 60 to 70% accuracy,
even with a 30% failure, will say that potentially 1 million
to 10 million in the annual revenue, or in terms of the annual
cost they could reduce. And so, machine lending has been, you guys heard the buzzword
of a machine lending? It’s not turned into AI, it’s kinda the same conglomerate
thing, kinda wrapped up into different technologies like
Deep Learning Networks. You’ll here this more and
more from Microsoft as well, it’s becoming more AI driven. And the first start is just getting
people in the BI world to just plugin some of the out of
the box motors, even the data. And also,
I’m gonna show you a demo today. An example of what I’m gonna
show you today is something Microsoft has really spend
some time on building is this thing called
Solution Templates. Has anyone actually heard of
Solution Templates before? So just one, maybe two. So, Solution Templates is
designed to provide an end to end BI solution. So you come in here into
your Power BI website, you say hey, I have CRM,
I have Sales Management, I don’t know what reports
I should be looking at. I don’t know what models I need
to plug in, I don’t know what I need to see, but I’ve got my
CRM opportunity data with me. Can you make me a really
compelling use case business scenario with that? I said the Solution Templates
is a wizard. Where you just fill in your
information like your CRM instance, your CRM password, your SQL if
you wanna store the data in SQL, you had the pipe SQL, so you can
do good demo and takes on it. And then we give you a Power BI
file that has pre-made dashboarder reports. These reports in the SQL structure
that we make are enterprise ready, so they use the best practices
that we have inside the Microsoft. And today,
what I’m gonna show you is, I’m gonna go through the campaign
and brand manager for Twitter. Just briefly outline
what that actually means, and I’m gonna show a news template
which hasn’t released yet. And so, before this session kicked
off, what I decided to do was go into the Power BI website and
actually spin this up for real. So, let me show you how
quickly you can spin one up. I’m not gonna go through
the whole process, but, just to give you an example. So, if you go into Solutions, you’ll see the thing
called Solution Templates. It’s a relatively new thing. It’s only been there since I think,
June from last year. And we’ve been releasing one
around about two to three months, we’ve been releasing. And so, what you can do is if
you have a customer you can say, click Learn More. And you can see a review report. And you can say hey, if you want
to do campaign brand management using Power BI, here’s an example
report what that would looks like. And you can actually go in,
what we did was here, we captured Power BI hashtags, I
said hey, let’s look at the tweets, and there’s a nice cool custom video
that I will show you in a second, which allows you to quickly
analyze the sentiment data. So, this is where I am trying
to bring in advance analytics, really simple sentiment analysis
that you’ve probably seen thousands of demos before. Now, bringing that inside Power BI. One of the other cool things is
your customer may say, hey, but my organization is Walmart,
or my organization is Tesco. Can you show me with my data, that’s a very good question
that they only ask? And the great thing is
within eight to ten minutes, I can actually spin this demo
up very quickly for myself. So, if I click Install Now, this
would take me to a provision page, where we spin up a range of idle
resources, and I click Next. It’ll ask me for my Azure
credentials, and it’ll ask me for my Twitter search terms and queries. What do I wanna search for
in Twitter? And then right at the end of the
process, I’ll get a Power BI file. So, anyone who has an Azure
subscription can go in today, spin this up in the back page in ten
minutes, you will have a Twitter report showing the tweets
that you’re concerned about. And you’ll be collecting
those in real time. So, as you keep refreshing your
power report, you’ll see that your tweets will just be keep coming
in, and you can see you can be monitoring Trump, Clinton,
if you were watching the debate. You can see those tweets coming in,
and see the different trends and clusters that formed with them. So, what I’m gonna show you is
actually did the same thing for MS Tech Summit this morning. I spun up one before
this tech summit. Let me just open up this Power BI
file, and see how it’s coming along. Now, I did not look at this file, so I have no idea whether there’s any
meaningful data, but let’s just see. If I go into
the [INAUDIBLE] Template. So, the [INAUDIBLE] Template
uses machine learning, uses clustering, and
uses sentiment analysis. Very simple pre-built models,
using the cognitive services. We use a Logic App inside Azure,
if you know what Logic App is. Basically, we use it as an ETO. So we say once we get a tweet in,
can you perform sentiment, and can you do topic clustering? And that’s it, and
then pipe it to SQL. It’s a very simple Logic App,
I’ll show you that later on, I’m just gonna refresh this. So, it’s connecting to SQL. And lets see if I can
see any clusters. So in PowerBI I showed that we have
custom visuals we have the whole gallery dedicated to custom visuals. The MSR team that made this
Microsoft Research time actually made these reports. And actually made this cool visual
called a clustering diagram I’m just gonna click. Or the hashtag and
it’s gonna zoom in, zoom out,
show you what this graph shows you. So in the center of this
graph you’ll see a black dot. It’s black is slightly greater. The black dot basically means is
what topics am I listening for? And in this case the hashtag that I
was listening for is MS Tech Summit. So it’s by natural cause that I
expect you guys to have tweeted MS Tech Summit in
most of your queries. Which is why I said the center
mass of this clustering diagram. The green dots are all the people
who have tweeted about it. So if anyone is a top tweeter here,
I probably can find you in here and see how you have responded
to some of these tweets. I’m not gonna do that because,
just in case, if there was a negative sentiment,
I do want to remain. And so,
the black dots are clusters so what I might be
interested in saying hey, let me see what other clusters I
can see that in the Tech Summit. And so I expect very similar
someone’s been talking about Office 365 there must have been some cool
keynote regarding the MS 365 Office 365. You guys tell me I wasn’t part of
the keynote this morning I presume there was some Office
365 announcement. What I can do is drill in
specifically to the keynote and say hey, what clusters
formed during the keynote? What was talked about? And it’s no surprise that
I see the keyword keynote, keynote Microsoft come up. I can also see Jason Zander. It appears that people were talking
a lot about Jason Zander’s keynote. And I can see some other
clusters as well like BI. What’s that partners,
intelligent apps. I presume that was an intelligent
apps demo during the keynote today. You guys can tell me
what went on there. And I can quickly understand
what happened at specific times. I can do the same for
tomorrow’s keynote or closing keynote if there is one. I can do the same for
each session and say hey, what are people talking about? What are people really enthusiastic
about in terms of Microsoft and the MS Tech Summit? Again, I’m gonna go to another view,
and I’m just gonna go to
pivoting overview. And I can see that on average, you guys are very positive
about the MS Tech Summit. This is great to see. I’ve got some people
talking about potentially, some very low negative scores,
so maybe, as neutral. And some people, who may have
speak in a different language, that wasn’t picked up by
the sentiment analysis. There’s an undefined
section there as well. And again,
I can click on specific tweets, I can say, hey,
let me look at the offers. The MS I’m not gonna pick
on any specific offer. I can see people
are talking about Chicago, because we’re based in Chicago. And again, I can quickly analyze the
tweets that are coming in real time. I wouldn’t normally sit there and
watch every single tweet, that would be impractical for
me to do. There could be thousands and
thousands of tweets. But in this case I can just
get a quick feel of Quick tweets I need to respond to. If there’s highly negative ones or highly positive ones I wanna make
sure I’m engaged with the community. Later on I’m gonna show you
the app story when Power BI and BI meets apps is. I could then set up a job which
says when I get a negative tweet can you go and act on it? Can you send a response message to
this user saying I am very sorry for the discomfort we have caused you? Let us get someone to help you with
his problem it could be someone unable to find a room. I find a tweet earlier on today, can’t find it now where someone
couldn’t find a room and they tweeted about and
MS Tech something got back to them. Some good examples where you want to
kind of automate the process flow. I said this is the twitter template
you guys can spend up today or so put your own hashtag and
queries in. And you’ll get the same
report I’m showing you here. There’s no magic,
there’s no specific demo material, I literally about a couple hours
before the keynote this morning. So I’m just gonna close this down
actually, and I’ll show where if you really want to go down to we’re
gonna have a new demo template. Where we bring in news articles, where if you give us
the news article name and your Azure credentials, we will
create you this news template. Which will use the cognitive service
library from the Azure gallery. Which uses sentiment analysis,
uses topic clustering, uses key phrase extraction, and uses another type of clustering
model, which I’m gonna show you. So in this case I4 RBB experimental. Let me look at Pokemon Go, this demo was done in October when
Pokemon Go was a bit more prevalent. So you can see the top corner,
you can see the data. And I’m just going to
walk you through this and I’ll show you how much power
you can act in business and tangents if you just add a bit
of machine learning in to it. And so on the far left bottom
corner I can see topics, topics that have come
from within the articles. And so if I just zoom in, I can see some very interesting
ones about Tech Mobile. I can see Donald Trump,
Clinton coming up as very popular. And I might be very interested in
why these people are coming up with Pokemon Go? How are these articles
somehow related? I might do that in a second. I’ve seen some other topics
where people told Hague, which is a place in the Netherlands. And I’m like why are these
places coming up and I acquire popular articles
as well in my topics? 24 people have mentioned
this topic in the articles. There are, some of the other things I can do is
I can look at, we extract entities. And entities like people,
organization, location. And so I have organized it in
terms of organization or location, I can just click View More. And actually view a list
of organizations I’ve been talking about. And I see Microsoft Research
is there as well. I’m like, wow! With that why is that? And I can actually click in and draw
down into specifically that more. I’ll do that in a second. One of the other things
I can do here is again I can view it by sentiment. So the tabs at the bottom I can go
to sentiment, I can go to topics. And what I gonna do for you guys
is now and I’m just gonna go into, let me see. Let me see why Microsoft Research
is prevalent here. So, again entering cross filters, I can see exactly people are talking
about Microsoft Research. Many from Tuesday the 11th, and
I think that’s in November. And I see this top out of all and
I am going to click on it. And I hopefully expand this. Now, one second. And, the headline is
Pokemon Go could add up to 2.83 million years to a user’s life. So Microsoft Research
did a study and, as you can see, the top trending
tweet or article, not tweet. The top trending keyword that went
alongside these articles were steps. So as you see on the top right I
could see the trending keywords just quickly getting this over with. The great thing here is I can also
see which entities appears here. So I can see Microsoft Research
is mentioned in red, I can see the locations
mentioned on this article. So I could just quickly navigate
this article without having to read it so I can said hey hey. I’m interested to see when
Microsoft’s Research is mentioned. So I might just click on
Microsoft Research and it’ll just navigate me
across the article. So in this case I can analyze
a very long article and quickly analyze which parts of
it am I really interested in. In this case Microsoft Research did
a good study where they actually determined that people who played
Pokemon Go are more likely to live longer because they
actually do more exercise. And so they worked out that 2.383
million years of lives could be saved just by erasing
the Pokemon GO app. And this is something you would
not normally have considered. There’s also one by NASA that
they’ve tried catching Pokemon GO in NASA. I’m not sure if you saw the article. In the space they were catching
Pokemon GO characters as well. And these are fun things, insights
you’re not aware we are able to do. This is another template we’re
going to be releasing very soon, hopefully at the end of this month. This uses a range of advanced
analytics capabilities. It uses a range of
cognitive services. And what you really require
is an Azure subscription and the news articles you’re
really concerned with. And we’ll spin this whole template
for you as you see it today. So I’m not gonna spend
too much time on it. I could sit here and go through
sentiments, seeing how people look at the Clinton tweets and go through
how the sentiment was there. I could start going
into the different topics across the articles. And again I’m just using
the Power BI capabilities, the BI capabilities. And then here I’m
using the full stack. I could be using SSAS. To power the tabular model I keep
using Power BI to actually do the visual front end. To do the analysis,
I’m using Azure Machine Learning. I’m using cognitive services. I’m using the full Microsoft
breath of the stack. But most people would
normally struggle with trying to integrate
Lava Stack together. That’s why the solutions templates
is a great example of that. You just come in Azure subscription,
give us the topics. We’ll go and do the work and
you actually can go and customize the MM1 afterwards. We leave on your subscription,
nothing’s a black box. So if you guys wanna had some
really sophisticated for your customer’s needs please
feel free to do that. The templates
are completely free for you to use, the only
consumption that you pay is for the Azure subscription cost, for
the extra underline services. These templates are 100% free and
will always remain that. So, I’m just gonna come out of here. And I what I’m going to do is
quickly going to the final segment which is apps. I don’t have too much time to
go into the apps demo, but I will quickly briefly
talk about it. So the final section
is the fourth wave and this is where I see people
really trying to figure out and realize that SSASifying applications
is such an important thing. BI is great, but if someone can’t act on the BI
it’s not really meaningful. They end up having to go through 15
other different portal pages in that organization. To actually carry out the task. And so what we’ve done as a first
step is actually release Power BI embedded. Which is a small
Power BI Azure service. Which it provides a list of
API’s which allows you to embed reports into applications. So the person that is now using
your application does not need a Power BI license. That’s the key thing here. You can embed it. You pay the Power BI
subscription cost. It’s done on,
it’s a different pricing model. If you want to find out more, just
go on the Azure Portal calculator and you can find out that. And I can quickly integrate Power BI
of, it’s really not Power BI. I can integrate visuals
inside my applications. And by integrating my visions inside
my application, I can now create a very rich application that
not only allows me to see, hey my cells are failing. Or there’s this item that’s failing
inside my aircraft that I need to repair if I’m doing IT solutions. I can then click create service
ticket or create action and email with my staff. And I can create act and we call it
the act orientation where I measure using analytics, but
I can act on these analytics and actually drive better
data driven decisions. And so Power BI Imbedded is kind of
the first way where we have a list of APIs, and you can use
a range of different APIs. We have the Azure Resource Manager
which is basically the Azure APIs along with the rest APIs for
Power BI. So I can say hey, give this
Power BI report to someone but render it in this way. I can say when someone slices
the report, navigate them to this page and let them see
the report in a different view. So if I’m using CRM and
I click on a graph and I say hey I wanna look at my top
accounts, look at the revenue and I say I wanna filter on these accounts
that have the lowest revenue. It can directly go and
show me those accounts. So I’m now creating
rich applications, but using Power BI as
a mechanism to do that. I had a demo to show you
which is just a browser page. I’ve just run out of time, I wanna
save some questions and answers. If you guys wanna get any
of the demo material, I’ll post it online
onto the blog post and you’ll be able to quickly go in and
see all the demos I’ve done today are readily available on
the Power BI website. I now open the floor for questions. I’ll be here until
the end of the session. Thank you guys.>>[APPLAUSE]
So.>>In the Power BI Embedded [INAUDIBLE]
>>So in the Power BI Embedded scenario, you only bind it to a SQL Server
store where the data comes from. And that will be your SAS-ed
version of SQL Server. That will be, I met,
I know the person, who creates the application and
manages that. The user doesn’t need to even care
about their user permissions and then you map their AD token to
the SQL Server password and username that you want them to use. So you can manage that whole story. So yeah it’s a very
platform focused play. Like it doesn’t come
without the box. You could use Power BI Visuals
from GitHub and actually just get the data from your
own wherever it comes from, and then render the visuals
the way you want it. So Power BI Visuals is open source. You can go in and use the same visuals that
you see in and actually use them in
your own applications. So maybe that’s what you mean more
there, you can just bind the data set onto JSON payload for
the visual and they’ll render. Any other questions? Yes.>>On the Power BI there is JSON, they adjusted visual
like R visual or invest the ones that
contain that [INAUDIBLE].>>Custom visuals are supported,
I’m not sure about R visuals. I’m not sure if RRs
are supported because in R, we have a dedicated
R server that runs all the R visions through the engine
that we host for Power BI for you. For Power BI Embedded,
we wouldn’t want you to go and spin up your own R environment. And maybe that capability
may be coming, I’ll let you follow up with
the PMC where there is. You’ll see that on the blog post, I’ll posts in the blog post
with more information. Any other questions? Okay, thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *