THE SQL Server Blog Spot on the Web

Welcome to SQLblog.com - The SQL Server blog spot on the web Sign in | |
in Search

SQLBI - Marco Russo

SQLBI is a blog dedicated to building Business Intelligence solutions with SQL Server.
You can follow me on Twitter: @marcorus

The State of Power BI #powerbi

The 2015 has been an amazing year for Power BI. One year ago, I was commenting on this blog the public preview of Power BI dashboards. A few days ago, Microsoft released the December update of Power BI Desktop and several updates of the Power BI service. In one year, we moved from a situation where Microsoft seemed (and actually was) too slow in the BI market, to a weekly (and sometimes daily) wave of announcements that are becoming hard to follow also for those enthusiast early adopters (but don’t misunderstand and read my lips: D O N ‘ T   S T O P   R E L E A S I N G   N E W   F E A T U R E S !!).

It’s clear that Microsoft is moving quickly to get the cloud market of BI services, which is still unexplored (after all, most of the companies keep their data on premises), but it is the only player providing a credible story of integration between cloud and on premises BI services (see “The Box is Back” and “Happy Together” pictures used at last PASS Summit). I finally see both strategy and execution working together. And execution is happening at the right speed, which has to be very fast nowadays.

There are a long number of features, improvements, technical details, and future directions that justify this analysis, but there is still a long road ahead before good comments become actual revenues and adoption in production. However, there are key changes I have seen in this year that strongly support a positive outlook for the entire Microsoft BI stack of technologies:

  • No more “by design” answer: first and foremost, the entire Microsoft BI team removed the “by design” answer for bug/suggestions submissions. The Power Query team started with this positive attitude of considering feedback as something to use for real. Now the entire Microsoft BI team listen to the community, consider feedback and prioritize features based on comments received. Don’t underestimate this point. This, and an increased transparency, is the foundation of the credibility that leverages the word of mouth and community support. Other Microsoft products don’t have this asset, and they should do something to get it. Microsoft is behaving as a startup here, and the results are visible. One year ago, this was a promise. Today, it is a fact.
  • Constant and continuous innovation: weekly and monthly releases creates a push to keep the software up-to-date. This impacts in particular Power BI Desktop, and backwards compatibility is no longer a constraint for innovation. There are pros and cons in this, but by now there are more goods than bads.
  • DataZen acquisition: I’ve seen many comments that were not convinced about this move, because it was creating confusion and overlaps with other Microsoft products. Now that Microsoft released SQL Server 2016 CTP 3.2 that includes the first version of DataZen (now named “mobile report type”) in Reporting Services, most of these considerations are over. However, I always thought that DataZen acquisition was an excellent strategic move. Microsoft bought a fast-growing company that was not quoted, including the technology and the development team in a single move. While most of the comments were about the effective value of the technology and the overlap between DataZen and Power BI mobile app, my appreciation was more about buying a company to remove it from the market, so that other possible big competitors wouldn’t have been a chance to buy it. Doing that when the company was mature enough to have a real customer base (and a real working product), but small enought to not be under the radar of many analysts, is an additional bonus. Great move.
  • Custom visuals in Power BI: disclaimer: I might be biased for having inspired the creation of Synoptic Panel. But I think that making the visual components an open source part of Power BI and enabling a large community to contribute to a public gallery is a super smart move. If the community contributes, this is good. But even in case the community would not respond, Microsoft opens a door to release part of software complementing its product (Power BI), without requiring the cost related to official release of software (which higher costs are internationalization and documentation). Think about the cost of releasing and supporting softare in hundreds of languages (think to Office) and you should quickly realize why this move is smart. Not to mention the number of opportunities that it opens to Microsoft partner, leveraging the existing ecosystem.
  • Adoption or “R”: Is the support of the “R” language really required for a tool like Power BI? Let me say an unpopular opinion: there are too many buzzwords in the BI arena, and it’s not the first time. It has been always the same, just changing the buzzwords. We started with DSS (Decision Support Systems), we recently heard about big data, I’ve seen that BA (business analytics) rocks more than BI (business intelligence), you cannot live without a data scientist, and yes, you need “R” in your company. The list could continue for pages. But is this really true? I am convinced that “R” is very good in certain domains, and at the same time I see it used (or even just considered) also when it is completely useless. But I am a consultant, I don’t sell licenses, I help customers building solutions that actually work, possibly reducing development cost and time. “R” is an option, but is not necessarily important or relevant in many scenarios. But can Microsoft ignore the important of buzzwords and trends in the market? Absolutely no. So adopting “R” for data sources and visualizations (see latest announcement for R Visuals in Power BI) is another great move that will help Microsoft sales (and this is true for all Microsoft partners, too).

I have seen in one year a multiple of what I have seen in the previous ten years. In reality, Microsoft is leveraging many assets that were already in-house (the Tabular engine, Power Pivot, Power Query, Power View, and many other Azure’s based services), which development required several years. But the weak point has always been the “last mile” of data visualization and presentation. Other players were years-ahead in this area. And I think they are still ahead of Microsoft. You might have a scenario that already fits well with Power BI. Maybe your scenario will be “Power BI” friendly in a few months. But the trend is clear and the improvement is continuous. So, when I think to the considerations I will do in 12 months, I am very positive and I am also scared about the number of new features I have to learn. Since this is what I (probably we) asked for many years, I cannot complain!

I always take a look at stock price of companies to check whether financial results and expected outcomes correspond to my point of view. Unfortunately, it’s not fair to compare Microsoft with other specialized companies that work only in the BI market. So we cannot draw any conclusion by observing that Microsoft (MSFT) is at its highest point since Tableau IPO (DATA), whereas Qlik (QLIK) didn’t add much value to stockholders in the last 3 years (source: http://www.nasdaq.com/symbol/msft/stock-comparison).

 image

What can we expect in 2016? I expect Power BI to continue its growing as data visualization platform, ecosystem, and adoption. However, the upcoming Microsoft SQL Server 2016 will be a huge release for the BI services. It will be the more important release since 2005, and this is mainly because of the new SQL Server Reporting Services (see Microsoft Business Intelligence reporting roadmap to read more). In reality, also SQL Server Analysis Services is a very important release, being the first major upgrade since SQL Server 2012 for Analysis Services Tabular (see latest announcements for SQL Server 2016 Analysis Services CTP 3.2).

Without going in detail, there are many new features that will make Tabular a more “enterprise ready” platform. It’s not that the current version is not good, I helped many companies in succesful adoption of Tabular, but there are many improvements in productivity and performance that completely justify the upgrade even before considering new features available in the new version. There will be a lot to write about that, and I really look forward to start doing that in a few weeks.

The magic key will be “Hybrid BI”. It’s already happening. I expect most if not all of the companies to move in an “hybrid” area for their BI solutions, where certain part are implemented on premises, and others in the cloud. And the reason to move on premises is not always preserving legacy investement, it could be requiring top performance of hardware for in-memory databases. For a certain size, you need to do an accurate hardware choice, where spending more is not necessarily better. In the very long term, I expect everything to move to the cloud, but the reality is that the two technologies will live together for a very long time, and Power BI is a good candidate to play a major role in this scenario.

Published Monday, December 21, 2015 4:35 PM by Marco Russo (SQLBI)

Comment Notification

If you would like to receive an email when updates are made to this post, please register here

Subscribe to this post's comments using RSS

Comments

 

Radim Hampel said:

December 21, 2015 4:42 PM
 

Marco Russo (SQLBI) said:

Great idea - thanks for the links!! :)

December 21, 2015 5:01 PM
 

David Laplante said:

Great analysis, as always!

Looking forward reading posts in 2016!

December 21, 2015 6:53 PM
 

Marco Russo (SQLBI) said:

Thanks!

December 21, 2015 7:29 PM
 

PedroCZ said:

Any tips if Mac support is even coming?

December 22, 2015 8:22 AM
 

Marco Russo (SQLBI) said:

I think that Mac support for Power BI Desktop is not a priority - you can access the service, and the designer is supposed to be the tool to create reports, just as Visual Studio (and there is no Visual Studio for Mac).

December 22, 2015 9:07 AM
 

Dung-Anh LE said:

just in love with this phrase "But I am a consultant, I don’t sell licenses, I help customers building solutions that actually work, possibly reducing development cost and time."

I had the same thought since what we called Big Data :)

And also, with your contribution (and your team) in "self-service" world, we make a big move in bringing "productive" solution to our customer. So a big thanks to you & your team ;)

My favorite quote : " Maturity is not when we start speaking big things…it is when we start understanding small things"

December 22, 2015 9:29 AM
 

Marco Russo (SQLBI) said:

Dung, your words are a nice reward for our job - thanks!

December 22, 2015 9:33 AM
 

Brian Mather said:

Great wrap up of a great year!

December 22, 2015 10:10 AM
 

Marco Russo (SQLBI) said:

Thanks Brian!

December 22, 2015 10:33 AM
 

_Teva said:

Hello Marco,

Thank you for this article.

I also have some concerns about R in PBI. I currently having mixed feeling about this.

I have mostly seen R being used in environments where you need to train a model on your computer on a subset of the data, then you send your trained predictor to the production environment to have it filter each and every transaction (as they come) in order to make an automated real time decision (for example : spam/not spam classification).

In the enterprise BI world data is often (unfortunatly) kept within silos, the DW is udpated every hour and the transaction already happenned in the OLTP systems. By the time we get the data in the DW it's already to late to make a decision &/or prevent anything to happen, which is for me the added value of Data Science.

That being said we now will be able to predict the budget, the sales, automate the forecasting, do a classification work without using the Pareto classification  (http://www.daxpatterns.com/abc-classification/) by using let's say the traditional K-means. I am even wondering if this can replace the M language for the ETL job for the folks (like me) who don't want to spend time learning M.

Time will tell.

T.

December 22, 2015 10:55 AM
 

Marco Russo (SQLBI) said:

R in Power BI is a language to prepare data (it becomes a data source) and to visualize data (custom visuals). I don't understand exactly what is your concern...

December 22, 2015 11:03 AM
 

Mark Ginnebaugh - DesignMind said:

Marco,

Great post.  I met with a Senior Program Manager for Power BI while in Redmond last month.  It was great to learn about the resources being invested into the platform.  I was amazed at the size of the team they have dedicated to this.  It's going to be an exciting 2016.

Continued thanks for your insights!

Mark

December 22, 2015 12:00 PM
 

Marco Russo (SQLBI) said:

Thanks Mark!

December 22, 2015 12:05 PM
 

_Teva said:

Well I was saying that to me the value of R lies in the fact that R :

- is not a functional language,

- has great (mathematical) libraries,

- is usefull when you want to write a formula that can find optimum values on a mathematical function (for your data set).

Which is important when I want to do inference or classification. So adding R in PBI is something interesting that will have my attention in the coming days. I also hope that this will help widen the usage and spread of PBI.

I also envision a scenario where I do the transformation using R (instead of "edit queries"), wich is good (to me at least).

However PBI is used to display valuable information to a data consumer in order to assist him in the decision making process, that's not quite the usual usage of R.

December 22, 2015 12:33 PM
 

Marco Russo (SQLBI) said:

I finally see a good example of using R - up to now, the number of real use cases I've seen was very low! :)

Thanks!

December 22, 2015 12:50 PM
 

Ryan Wade said:

I personally think adding the ability to create R visuals is a good thing and is a worthwhile addition to me. Being able to use packages like ggplot2 is a much easier way for me to create custom visuals then learning typescript. Ggplot2 is based on the "grammar of graphics" methodology which is very intuitive to me and something that I already know. I think the type of custom visuals that I will need I should be able to create using R ggplot2. I have a custom visual that I plan to develop soon using ggmap and I hope it works!

Thanks for the post and the knowledge you openly share with the community!

December 22, 2015 5:01 PM
 

Marco Russo (SQLBI) said:

THanks for your feedback Ryan - another good testimonial for R! :)

December 22, 2015 5:14 PM
 

Brian Mather said:

Teva Thanks for your insight, am going to spend some more time with R. Have you any good resources to recommend as a start?

I felt the same way about M, but I can now highly recommended learning the basics. I've not invested much time so far but every moment I've spent has saved me in spades. It really does reward you early on in the learning curve. Chris Webb's blog is fantastic, an about to buy his book.

December 23, 2015 3:42 AM
 

_Teva said:

Hello Brian,

It depends, my point of view is if you need to do descriptive statistics, Dax is your friend, I would more than recommend Marco's book & blog (http://www.daxpatterns.com/statistical-patterns/).

If you need the visualisations or to performs specifics operation (egg : find an optimum value regarding some given constraint [as solver do in Excel], find the eigenvectors on your fact table) then you want to check cran's website. R's help and IntelliSense are very good but last time I checked was not available (yet?) through PBI.

December 23, 2015 6:50 AM
 

Jason Russell said:

Marco,

Thanks for the post, and for the Synoptic panel! Looking forward to what SQLBI will come up with in 2016.

I'm most excited about the fact that Power BI is a powerful tool for those of us who aren't developers. The tool is user friendly enough that with limited time spent learning, "normal" people can create powerful visuals to share insights.

December 23, 2015 2:17 PM
 

Marco Russo (SQLBI) said:

Thanks Jason (and yes, we're already working on two or three things for 2016, there's a long road ahead!)

December 23, 2015 6:00 PM
 

Francesco Civardi said:

Hi Marco,

Another example about using R in Power BI: Import Salesforce tables in Power BI, typically the Opportunity table, train a decision tree with R, which identifies the best attributes (and their values) for a "conversion", and output the tree, without leaving the Power BI environment.

I love using Power Query and M to "munge" the data (while I hate using R for that), and I love letting R do the number crunching part of the Job. I think the Power BI - R integration makes "advanced data exploration" possible in Power BI (i.e. clustering, classification, survival analysis etc.)

December 25, 2015 11:25 AM
 

Marco Russo (SQLBI) said:

Hi Francesco, nice to hear from you!

Thanks for the feedback - based on your and others, it seems clear to me that R is for statistics what M is for ETL, and DAX for "business logic" within a data model.

December 26, 2015 3:55 AM
 

Tarek Demiati said:

Marco : M,R,DAX ... sounds like a huge alphabet soups that could scare the mere mortal to start a career in Business Intelligence ;-)

// Tarek

December 29, 2015 5:34 PM
 

Marco Russo (SQLBI) said:

Hi Tarek - I know, but that's life, we live in a complex world! :)

December 30, 2015 1:58 AM

Leave a Comment

(required) 
(required) 
Submit

About Marco Russo (SQLBI)

Marco Russo is a consultant, writer and trainer specialized in Business Intelligence with Microsoft technologies. He runs the SQLBI.COM website, which is dedicated to distribute resources useful for BI developers, like Integration Services components, Analysis Services models, tools, technical information and so on. Marco is certified as MCT, MCDBA, MCSD.NET, MCSA, MCSE+I.

This Blog

Syndication

Archives

Privacy Statement