Power BI (cloud) + SSAS Cubes - sql-server

1) Can Power BI (online/cloud-based) use our local SSAS cube directly as a data source?
2) If no, and I assume it is no, then can we upload our SSAS cube(s) to be used as a data source, and how do we do that, preferably incrementally (if it is possible to do that incrementally)?
3) If SSAS cubes cannot be used, then I assume that we have to use data built into the SSAS Tabular Model, and use DAX to query it?
4) If this is true, then how do we send data to there? Do we have to define the tabular model locally and ship the stored results (since the tabular model is in memory, I’m not sure that that even makes sense), or do we send constituent tables to the cloud and build the tabular model structures there.
5) If I build this in an SSIS package (which I gather I do), is it an SSIS package that is built and maintained locally (meaning on our existing database, running MSSQL 2012 w/ Analysis Services, the way our existing SSIS packages are), or is it built and maintained in the Power BI Online environment in the cloud?
We're looking at using the PowerBI Preview to deploy dashboards and scorecards based on data that we collect on-premises. I'm assuming that we'd use that OData plugin to make data available in the cloud, for starters...?
edit: thanks for reading!

Regarding 1): Power BI for O365 cannot as yet connect to an SSAS tabular or multidimensional model directly as a source. However, the new Power BI Preview (released December 14) does allow a direct connection to an SSAS tabular model with the new connector. So it is very likely that Microsoft will soon release a similar connector for SSAS multidimensional, as they did for power view on SharePoint, first releasing tabular then multidimensional.
2): There is no direct connection to user-maintained SSAS multidimensional models no matter where they are stored. There is a direct connection to tabular models with Power BI Preview, whether the tabular model is on-prem or in azure.
3), 4) & 5): If the purpose is to "deploy dashboards and scorecards based on data that we collect on-premises" then take a close look at Power Query and Power Pivot and at the Power BI Designer. Also take a look at the developer tools that are available here: https://msdn.microsoft.com/powerbi/

This is an old question, and things have changed between then and now. In today's context using Power BI Gateway and Power BI desktop you can connect your on-premise SSAS cubes to power BI cloud and schedule your cubes to refreshed automatically. Earlier it used to support only tabular model, but as of now it supports both tabular and multidimensional SSAS cubes. The only thing which is not yet available is you cannot live connect a multidimensional cube but you can schedule it to be auto refreshed. However with SSAS tabular you can live connect to Power BI.

We do this for a number of customers... you first need to consider the volume of data in your models as this will dictate your requirement for using Analysis Services (Tabular, as quite righty pointed out Multidimensional is not an option as the in memory model of SSAS Tabular is suited to PBI).
Then you need to think about how the dashboard will be updated and how automated this process can be made using Power Query + Power Pivot - using your OneDrive for Business could be an option to make life simpler should you not be able to use SSAS Connector in the preview.
Finally, depending on your source OData and a few other connections can be automatically refreshed on a schedule (CRM for example)...
Hope that helps.

Related

SSRS and PowerBI Server

I've been using SSRS 2012 for a while now. Keep in mind I'm currently using SSRS 2012 but have set up a 2016 server and will be migrating about 200 reports within the next few months. Just went to PBI training and found out about the new Power BI Server that can sit on top of SSRS. Exciting in that we're in healthcare and cannot use the PBI publishing service for HIPAA reasons. But, I wanted to be sure I understand some things:
In SSRS, you can create a datasource and datasets that are used
regularly for efficiency and to keep down storage sizes. In
PBIRServer, it appears that you create each datasource and the
individual datasets used and store separately for each report. Is
this accurate and doesn't that seem like a step back?
Can I include SSRS reports and BPI reports/dashboards on the same
site?
If we're going to set up a local PBIRServer, can we develop using
PBIpro with about 5-10 pro users but then let the folks that
basically just want to view data use the free version?
If we develop using PBIpro can we still publish to the PBIRServer
with mobile formats? Documentation seems to indicate we need a
different development tool with a much higher cost.
Can you include a hyperlink from PBIRServer reports/dashboards that
to a specific report on the same server? I’m seeing this being used
via PBI for the visuals and then the drill-down-to as the existing
SSRS reports. They’re working great for our current purposes.
Is there a publication that articulates some of these specifics?
Thanks so much!
I think the first thing to keep in mind is that reportserver 2016 and power bi reportserver 2016 are different products. Licensing Power BI reportserver can only be obtained by either buying power bi premium capacity or have an enterprise sql server with Software Assurance
PBI premium: Costprice for this will be 5000$ a month
power bi price calculator
SQL Server Enterprise: $14,256 per corepack , 2 are required + SA
I can't answer all other question, but for question 2:
Yes you can deploy power bi and regular reports to a pbiRS server.
Question 3:
When you develop locally you have to use the power bi desktop for reporting services. To deploy this to a pbi RS you are not required to have a pbi pro license. Since you are using on premise resources, you will follow the licensing model of sql reportserver. The users connecting to the reportserver are no power bi users, just regular ssrs consumers install power bi desktop for report services
If I understand your questions well, you might need to install both, depending on organization size, report creators number and report users number.
SSRS for those people who are OK using standard reports only (with exposed datasources and standard layout design tool) so SSRS yes included with your SQL Server license
Power BI Report Server (SQL Enterprise+Assurance or PBI Premium license) for more sophisticated reports for business people; but to design/publish these reports you need Power BI Pro licence, per report developer

Logical method of using SSIS to transform data and upload it to Azure Data Warehouse

My current model looks like this:
Gather disparate data sources and import into SQL Server.
Process and transform data using SSIS packages.
final step in the SSIS package uploads data to the data warehouse.
BI tools pull data from the data warehouse for end users.
Is this a logical work flow? I initially was going to use data factory and the Azure SSIS integration runtime to process data. However I didn't understand why these steps were needed, as it would seem simpler in my situation just to build my SSIS packages on premises and upload the processed data to my data warehouse. What benefits would I gain from using data factory and the integration runtime? My main concern is that my current model will make automation difficult but I'm not entirely sure. Any help is appreciated.
Your possible paths here would be SSIS on prem, SSIS on VM in Cloud, SSIS in ADF or natively build the pipelines in ADF.
ADF is an Azure Cloud PaaS managed service for data movement and data integration orchestration. To reach back into on-prem data sources, you need to use an Integration Runtime gateway on the source side. So, if you are looking to move to a Cloud-first architecture or migrating into Azure, ADF is a good solution (use V2).
If you are remaining all on-prem SSIS on-prem is the best scenario.
If this is hybrid, where you will continue to have some data on prem and load Azure Data Warehouse in the Cloud, then you can still use SSIS on prem with connectors into ADW as the target. Or if you have to eliminate the local server concept, you can run that SSIS in a VM in Azure.
If you want to eliminate both the datacenter server and the need to patch, maintain, etc. the SSIS server, then use SSIS in ADF, which provides SSIS as a Service. In that case, you can still move data in a hybrid manner.
It really is going to depend on factors such as are you comfortable more in Visual Studio to develop SSIS jobs or do you want to build the pipelines in JSON in ADF? Do you have a plan or a need to move to Cloud? Do you want to move to a Cloud-Managed service (i.e. ADF V2)?
I hope that helps!!

Does Process Cube eliminate the need for SSIS?

I am trying to understand how SQL Server Analysis Services fits into the Business Intelligence field.
I have used SSIS to create copy databases and then SSRS to produce reports, which are accessed by he users.
I know that SSAS is a database engine, which allows you to create Cubes. There is an option in SSAS to process cube (http://technet.microsoft.com/en-us/library/aa216366(v=sql.80).aspx). Is SSAS a replacement for SSIS as it seems to do the ETL for you (using process cube)?
SSIS is an ETL tool providing you with the ability to move, manipulate and consolidate (from multiple sources) data. SSIS tends to be a developer tool used to get the data in the correct shape either for an application or a reporting tool.
SSAS is a cube building tool providing the business with the ability to slice and dice the data ad-hocly. Developers will build cubes, however the consumers will tend to by the business.
I have seen instances of SSAS cubes built pulling data directly from source, but these tend not to work very well, due to the load on the source systems and the complexity involved in structuring the data correctly.
A more typical approach is to utilise SSIS to pull (possibly only daily differences) and stage the data into a dimensional model that can then be cleanly consumed by SSAS. This way both tools are playing to their strengths - SSIS moves the data around and SSAS presents the data in an efficient and user friendly way.

Reporting in TFS 2013

I am currently working on TFS 2013 and would like to know how to generate custom reports.
Tried google to find out a tutorial but couldn't find anything great. If someone has then please let me know.
Whether SQL Server 2008 R2 Enterprise Reporting Service will be ok for TFS 2013 to generate reports?
TFS has a few mechanisms for doing reporting:
Work Item Queries (WIQ) - Work Item Queries can be created and saved for reuse. Work Item Queries are easy to create, and can answer the majority of questions that users have.
Excel Reporting - Work Item Queries can be exported to Excel to provide more advanced capabilities around analyzing the data resulting from the Query. Various different views of the data along with charts/graphs can be created using standard Excel functionality. Creating Excel Reports requires some expertise in Excel, but the resulting file can be saved and the data can be automatically refreshed from TFS.
SSRS (SQL Server Reporting Services) - This is the most advanced Reporting option. SSRS reports require a developer to create, but it provides the flexibility to report and analyze any data stored in TFS in almost any way imaginable.
TFS Web Access (TWA) - TFS Web Access provides some basic charts/graphs that can be viewed in the browser (Burndown, Cumulative Flow, Velocity Chart, etc). In addition, web-based Charts/Graphs can be generated based on the results of Work Item Queries. This is less flexible than the Excel option, but also more user-friendly and usable by any TFS user.
The advanced option - SSRS Reports - are just your run-of-the-mill SSRS reports that use the TFS Data Warehouse and/or the TFS Cube as a data source. You can read a guide about developing custom SSRS reports for TFS here: http://vsarreportguide.codeplex.com/

Analysis Services with excel as front end - is it possible to get the nicer UI that powerpivot provides

I have been looking into PowerPivot and concluded that for "self service BI" and ahoc buidling of cubes it has its uses. In particular I like the enhanced UI that you get from using PowerPivot rather than just using a PivotTable hooked up to an analysis services datasource.
However it seems that hooking up PowerPivot to an existing analysis services cube is not a solution for "organisational BI". It is not always desireable to suck millions of rows into excel at once and the interface between PowerPivot and analysis services is very poor in my book.
Hence the question is can an existing analysis services solution get the enhanced ui features that power pivot brings, without using powerpivot as the design tool? If powerpivot is aimed at self service/personal BI then it seems bizare that the UI for this is better than for bigger/more costly analysis services solutions.
Although I agree that PowerPivot has a nicer UI than using Analysis Services via standard pivot tables, PowerPivot through the Excel client has some really bad drawbacks when trying to use it in lieu of Analysis Services.
You have to download all the rows into your spreadsheet to "refresh" the data. In large data warehouses, this is equivalent to having users run SELECT * queries directly against your database. It's horribly slow for the user and has a high resource usage cost to your server.
It is extremely easy for someone to either intentionally or unintentionally walk out of the office with your entire data warehouse in a non-secure manner. Ouch!
The end-user machines need to be pretty powerful. I tried using PowerPivot with a few small tables (5 million rows or less) on our standard company machine build and it did not have sufficient memory to refresh PowerPivot. The only way I can see to deploy PowerPivot across the enterprise is to upgrade all of the analyst machines to 64-bit Windows 7 with at least 6GB to 8GB of RAM. Although this can be feasible in a small organization, it is not a reasonable solution in a large enterprise.
You won't have any good metrics on how people are using your data if you hand out PowerPivot with unrestricted access to your data warehouse. Yes, you may have metrics on how frequently people hit the refresh button and you may be able to log which tables they are querying, but you won't see how they use the data unless you audit their spreadsheets directly. And even then, you will only get their final result -- not their path to how they got to the final result.
PowerPivot generates really, really big files. Even if someone drills the data down to a small subset of the total data, it is still difficult to share the files with others since large PowerPivot files generally exceed minimum Exchange server file size limits. I've encountered this at my organization despite never having had this problem with an Analysis Services files.
PowerPivot does not have a very good security model. Sure, you can restrict who gets to the data the first time, but you can't restrict it once it is in the spreadsheet. Analysis Services prevents users from making changes to the spreadsheet if they don't have access to the underlying cube. It's just so easy to compromise the security of your most valuable business data with PowerPivot.
PowerPivot does not currently scale for very large data sources. I have several multi-billion fact tables that just can't be downloaded by PowerPivot unless I pre-aggregate them down to a few hundred million rows. PowerPivot works really well for small data warehouses, but it doesn't elegantly scale to large data warehouses.
Please note my above comments don't apply to PowerPivot via SharePoint. I haven't tried the SharePoint integrated product out, but many of the above concerns seem to have been addressed from the documentation and demonstrations that I've seen of the SharePoint version of the product.
Despite all of the above comments, PowerPivot could work as a replacement for Analysis Services if you have a very small or immature data warehouse. If your largest fact table is a few million rows, then the overhead of building and maintaining a data warehouse may not be cost effective if you are a BI team of 1-2 people. PowerPivot is probably a great new feature for a department that doesn't have a dedicated BI team and only has a handful of Excel junky analysts. It doesn't take much sophistication to put together a virtual data mart from disparate data sources with PowerPivot. But if you want to build a truly professional data warehouse that is secure, scalable, and highly manageable, then I would recommend building cubes in Analysis Services and either use Excel or a 3rd party vendors tools for connecting to the Analysis Services OLAP cubes.
Now that SQL Server 2012 is released, you may want to take a look at using one or more SSAS BISM models, rather than PowerPivot. You get interop with PowerPivot, but you can now build your model using SSDT (in Visual Studio) and can get more control over security and can host on a dedicated server.
I'll be presenting live and online this spring and summer of the BISM - here's my latest deck on slideshare - http://www.slideshare.net/lynnlangit/sql-2012-bism
Now that Office 2013 preview is out, you can check out PowerView inside of Excel (PowerPivot) without the need to have SharePoint. It remains to be seen when MSFT will remove the dependency on Silverlight (i.e. move to HTML5). The preview release of Office 2013 that I got in September still included PowerPivot which required Silverlight. I am looking forward to the release built on HTML5. Here's a deck by Jen Underwood to give you an idea of what PowerView looks like.
WebPivotTable is a pure javascript pivot table and pivot chart component which can be used to pivot csv data and all kinds of OLAP cubes, include microsoft SSAS. It mimics all functionalities of Excel but web based, no dependence on any other plugins, drivers, server side compoenents. It can be easily to integrated into any web application and web sites.
Here is Demo and Documents
I know that Powerpivot is a free download for Excel 2010, but for a better desktop client experience you should look at the ProClarity client.
Also worth looking at Analyzer by Strategy Companion (http://www.strategycompanion.com).
I've found it provides a smooth web-based interface for slicing and dicing in pivot tables (and charts) that is nicer than what is provided by Excel 2007.
ProClarity was the runaway best option until Microsoft bought them and killed the product. Some of the features are making their way into other tools, but the product itself is no longer supported. Panorama or Tableau are probably the best 3rd party options.
This is the best I've found so far that is up-to-date: http://www.varigence.com/products/vivid/videos
Edit: http://silverlight.galantis.com is also a possible solution - WPF version comes out next month that could be used is a VSTO add-in.

Resources