The Codex of Business Writing Software for Real-World Solutions 2.pptx
SQL 2012 PowerView Talk SPSLA
1. Business Intelligence:
SQL Server 2012 PowerView
Ravi Bodla
Senior Manager
Hitachi Consulting
Orange County, California,
United States
2. About Ravi Bodla
• Senior Manager
– Microsoft Platform
– Southern California Region
• Hitachi Consulting
– Global IT and Management Consulting Firm
– Over 5000 consultants worldwide
– Microsoft Worldwide Partner,
– Microsoft Product Development,
– Partner Advisory Council
• Microsoft V-TSP
3. Session Objectives and Takeaways
• What is PowerView?
• Learn about PowerView’s key capabilities
• Architecture and requirements
• Understanding of the SQL Server BI Stack
• Have fun with data!
5. SQL Server2012
Delivering business agility and innovation to gain strategic
value out of your information
MISSION CRITICAL DEVELOPER & IT PERVASIVE INSIGHT
PLATFORM PRODUCTIVITY
Microsoft Confidential—Preliminary Information Subject to Change
6. SQL Server 2012 Reporting and Alerting
Empower users Increase Proactive Increase efficiency
Intelligence
• End User Alerting
• PowerView • Enabled as SharePoint
• Defined from within Shared Service
• Highly visual design operational or ad-hoc
experience reports • Built-in scale-out for
RS Service Apps
• Rich metadata- • Intuitive Alert rules
driven interactivity • SharePoint Cross-
• Alerts self-managed
farm reporting
• Presentation-ready through SharePoint
at all times • XLS/Word 2007/2010 • Integrated backup &
recovery, ULS
• BIDS integrated with logging, PowerShell
Dev10 shell etc.
Managed Self Service BI – Corporate BI
7.
8. What is PowerView?
PowerView is an interactive
data exploration and visual
presentation experience.
9. PowerView Highly Visual Design Experience
• Interactive, web-based authoring and sharing of information
• Familiar Microsoft Office design patterns
• Powerful data layout with banding, callout and small multiples
visualizations
Rich metadata-driven interactivity
• Fully integrated with PowerPivot
• Drive greater insight through smart and powerful querying
• Zero configuration highlighting and filtering
• Animated trending and comparisons
Presentation-ready at all times
• Interactive Presentation turns pervasive information into persuasive
information
• Deliver and collaborate through SharePoint
• Full screen presentation mode for interactive boardroom session
11. PowerView Architecture
SharePoint Farm
Web Front End App Server BIDS
Tabular Model
RS Server
RS Web DAX
service extension
PowerView client ADOMD.N BISM File
ET (optional) AS
Server
Tabular
mode
(Optional)
DirectQuery
Web Front End PowerPivot
System Service
Excel PowerPivot PowerPivot
Model web service AS Server
Sharepoint
Integrated
SQL Server
12. Analysis Services: Tomorrow
Build on the strengths
Embrace the relational
and success of Analysis
data model – well
Services and expand its
understood by
reach to a much
developers and IT Pros
broader user base
Bring together the
relational and Provide flexibility in the
multidimensional platform to suit the
models under a single diverse needs of BI
unified BI platform – applications
best of both worlds!
13. BI Semantic Model: Benefits
Flexibility Richness Scalability
o Tabular and o Serves entire range of BI o VertiPaq in-memory
multidimensional engine
modeling solutions
o Cached or passthrough o Rich modeling o State-of-the-art
storage capabilities compression algorithms
o VertiPaq for o Scales to largest enterprise
performance, MOLAP for o Complex business logic servers
scale
o Fine-grained security o Improved SharePoint
o Choice of end-user BI
tools configuration and
performance
14. BI Semantic Model: Architecture
Third-party Reporting SharePoint
applications Services Excel PowerPivot Insights
BI Semantic Model
Multi- Tabular
Data model dimensional
Business MDX DAX
logic/queries
Data access ROLAP MOLAPVertiPaq Direct
Query
Databases LOB Applications Files OData Feeds Cloud Services
15. Data Model
Tabular Multidimensional
Familiar model, easier to Sophisticated model, higher learning
build, faster time curve
to solution
Advanced concepts baked into the
Advanced concepts (parent- model
child, many-to-many) not available and optimized (parent-child, many-
natively in the model… need to-many, attribute relationships, key
calculations to simulate these vs. name, etc.)
Easy to wrap a model over a raw Ideally suited for OLAP type apps
database (e.g.
or warehouse for reporting & planning, budgeting, forecasting)
analytics that need the power of the
multidimensional model
16. Business Logic
DAX MDX
Based on Excel formulas and Based on understanding of
relational concepts – easy to get multidimensional concepts – higher
started initial learning curve
Complex solutions require steeper Complex solutions require steeper
learning curve – row/filter learning curve –
context, Calculate, etc. CurrentMember, overwrite
semantics, etc.
Calculated columns enable new
scenarios, however no named sets Ideally suited for apps that need the
or calc members power of multidimensional
calculations –
scopes, assignments, calc members
17. PowerView is NOT
Crescent is an interactive data exploration and visual
presentation experience.
• Does not replace RB 2.0, 3.0 or BIDS
• Not a goal to edit or add new interactivity to
Dev/IT Pro reports built in RB or BIDS
• Not a high-end analysis experience
– Not a goal to provide complex calculation building
• Not a cell-based calculation tool
• Not a forecasting/write back tool
• Not a replacement for PPS scorecards or ProClarity
18. Report Designer Report Builder Project “Crescent”
Embedded Operational Business
21. We want your feedback!
Use this QR code or visit:
http://sps.la/feedback
Silver Sponsors:
22. Victory Lap- social event
"SharePoint Victory Lap" Social Event for
SPSLA will be at: 5:30pm to 8pm at
Di Piazzas (5205 E. Pacific Coast Hwy, 90804)
Hinweis der Redaktion
Where do we want to take this product. Multi-release, multi-year vision for the productSometimes you want fast time to solution, sometimes you want complex calculations and huge scaleBISM is the name we give to Analysis Services that encompasses these goalsSession Talk------------------------now when we started SQL server Denali, at the beginning of the planning stages we say let’s take a step back, where do we want to take this product not just for Denali, but for multiple releases. Let’s come up with vision for this product where it take Microsoft BI, and come up with multi release vision that we can articulate and then deliver on over a period of multiple releases with Denali being the first release. According to the coordination of AS the --familiar with sort of conceived almost a year back and we were studying --vision and we target a good time to takes talk --sorts of applications customers are building or want to build and we want to come up with a vision that captures those needs.So first of all we said we have got a very rich and large ecosystem, developers, partners IPSVs, we want to carry this ecosystem forward in a seamless manner we don’t want the community to learn start things from scratch, community to build the top of work they already have, at the same time we want to extend the reach of Microsoft BI, so if you go the typical IT conference let’s like Microsoft Tech-head and you can ask around what do you think about BI, you can get some people would say hmm, I have heard about it but not really sure about that what it is, you can hear few people say I have heard all up about multi-dimension modeling and BI can do the legal things but it seems like something that’s complex, high learning curve not sure how to get time to go about getting into that.And on the other hand if you go asking about traditional relational database concepts of SQL you are going to find people are much more receptive to that, there is a certain amount of penetration a certain amount of familiarity with relational database concepts that exists out there and we want to sell of embrace that in BI and in AS. This is where we want this relational data modeling because in order to --more IT professionals and developers that are familiar with that, at the same time we recognize that what AS have today with multi-dimensional modeling, with level of richness and scale on the --- so we want to keep that an essential that provides the best of both ways. We recognize that BI applications don’t come in single shaper size they have varying needs; some applications need tremendous requirements in terms of scales and performance, some application have a lot of needs in terms of calculation complexity, there are lot of BI applications that need rapid attritions, there is a lot of time hear from customers coming and saying my usage have a need and for me to go provision fund project --go through the --process of ETL and --development report development, all of that takes a long time and I’d like to be able to iterate really fast, and some of these things. So that’s another requirement that we hear from --we need we would like this platform to be able to provide the flexibility that it can meet all of these needs, based on the needs of your applications. So kind of the BI Semantic model is it on we use, to capture this vision, the vision that takes into account all of these needs or all of these principles that we have outlined here.
Carry forward the existing ecosystem of client applicationsLook a bit deeper into the conceptual architectureCall out PowerPivot as a clientMDX versus DAXData access: cache the data or pass it through Vertipaq new in-memorycolunmstore, compression, blazing fast perf. No aggregations or tuning. Same as PowerPivot for Excel and SharePoint.Session Talk------------------Let’s just get into the details of the model, so here is a conceptual architecture of the BI Semantic model, at the bottom you see a list of data sources. The model can consume data from a variety of data services; you have got classic relational databases, SQL server, Oracle data unit, all of the traditional relational databases ---pull in data from all data fields in SharePoint list, for example ----you can pull data out of those, you can pull data from Facebook and in a Netflix if you want, anything all of the public data services that are exposed by all data, you can consume SQL --report data service. So, the idea is that we want to make sure that pretty much you got data there in any form, even in text file or excel workbook sitting on a desktop, you should be able to consume that into your model.And then finally the model does a bunch of transformation with the data and serves a top to variety of client tools, the client tools in Microsoft BI stack are listed there if got reporting services, for both professional and operational reporting as well as ad-hoc reporting with crescent, you have got excel, you have got SharePoint inside which is the --for the BI capabilities in SharePoint which is performance points scorecards and dashboards and so-forth. The analysis service continues to expose a rich set of public API for client tools to consume those already from a rich ecosystem of client tools out there. They are not built by Microsoft and you have got tools like ---a number of tools and all of these tools continue to be supported and so we expect that there will be a significant number of third party tools built on analysis services as well. And then you have got PowerPivot as well which is very interesting, right? This is kind of take a little bit of time to think about it, PowerPivot itself has a semantic model embedded inside it, the user doesn’t know about it, the user is just doing his job in excel and then a model gets --out but that model inside PowerPivot can also consume data from professional model that is available on analysis services server so you can essentially let an IT put out an application which consist of BI semantic model, on an AS server, then an information worker can then launch PowerPivot may be take a portion of that data in model and then mashed up with some other personal data that they might have and then build essentially a new model, so that’s kind of builds up interesting scenarios as well.So, let’s look at inside the model, as what is in the model, so we like to think of the model in terms of three layers, the data model layer, the business logic and query layer and the data access layer. So walking from the bottom up, the data access layer is essentially the layer that consolidates the data from the variety of data sources that you have and then --to the higher level in this stack, now the data can be cashed, or the data can be passed through. Cashing the data in analysis services knows how to optimize the data, compress the data and aggregate the data as needed, and essentially make it available in a form that interacting with the data comes really-really fast. That’s can a high your usage get that response time at the speed of top, and we have got them all up engine, this is the traditional engine for cashing in analysis services, and we’ve got a brand new in memory --engine called the work pack engine, that is introduced in Denali, that enables cashing as well and then you have got a couple of different ways you have got roll up and a new mode called direct query. That enables you to pass through the access to the data, so you have got an application where you absolutely don’t want to cash the data, you want the queries to be pass through to the back end data source, and you can do that as well. It’s traditionally possible with roll-up, with some limitation and direct query is a new mode that we are introducing that will also let you pass through the access to the data, and we’ll cover that in more details. Business logic, once you have pulled in the data then you have decided whether you want to may be cash the data so that your users get fast performance, fast access to the data and one of the big elements of being data modeling in analysis services is to enrich the model because this model is the one place where all of your business rules are defined. If you have got sales model that is and you have a business matrices that defines how your sales are doing, your -related to previous periods, may be you have certain KPI or matric that defines your measure of success or failure with regards to some business goals, well, you want all of the end users regardless to what will be the reason, to see the same central definition of those matrices, or business rules, and analysis services offered with MDX and DAX --expressional languages to define those rules. And our MDX is being with AS for several releases so it cannot become the defective standard in the industry, you’ll for multi-dimension data access, but with PowerPivot we introduced DAX- data analysis expressions, which is an expression language sort of based on excel formulas, the goal there was to reduce the barrier to entry we’ve acknowledge that MDX while extremely rich and sophisticated powerful the barrier to entry buzz exist there to learning curve involved with MDX and with DAX we try to reduce that barrier to entry, the idea is with information workers and professionals if you are familiar with relational database concepts, tabular concepts, tables and columns and relationship navigation the barrier to entry with DAX is going to be much lower. And there is an entire session devoted to DAX, either today or tomorrow and so I’d encourage you to go and attend that one. And then finally we’ve got the data model, now the data model is to look to at it, one is that the data model is a model that developer works with so the model developer can work with the multi-dimensional model when building a model or work with a tabular interface.Now you guys use AS today, so you are familiar with the multi-dimensional paradigm of major groups of dimensions that what john was showing you the demo was a completely different paradigm it was just the tables and relationships interface to building models, so you have both options and regardless of which option you use the client tool can then choose to consume the model in either of the two ways, so if the client tool happen to be excel, excel just consumes the model as a multi-dimensional model because excel was built and that multi-dimensional model is more conducive to interactive analysis but crescent chooses to consume it using the tabular interface. Okay, so one of the points that I want to make here is, this model is a hybrid model, it’s a blend of both relational and multi-dimensional modeling