引子
和大家一樣，我們想聽聽大師來聊聊關于Power BI的事，那今天這篇要好好讀讀咯。
本文來自2017年9月新晉微軟Power BI方面MVP德國伙伴Lars Schreibe與Marco Russo的對話記錄。該對話記錄由Lars Schreibe整理于自己的博客（https://ssbi-blog.de/ 注意：他是德國人，博客是德文的，但本文在文章后部有英文記錄），其內容為Lars了解Marco的心路歷程。
相信對Power BI感興趣的你一定有很多方向和職業(yè)的問題想請教大師，那Marco可以算是該領域的國際級大師了，如果你不知道Marco Russo，你可能聽過兩個“意大利人”教程以及“綠皮書”教程，沒錯，這些都是Macro的貢獻。
本文翻譯一下年輕MVP Lars 與 Marco的對話記錄，其實有點像是采訪，但其內容是很有意思的。
原文地址：https://ssbi-blog.de/power-bi-journeys/die-persoenliche-power-bi-journey-von-marco-russo-in-english-as-well/
我們將了解到以下重要的有意義的話題：
- Power BI的歷史
- 關于快速度量值你怎么看
- 初學者秘笈
- 到底選云還是本地
- OKViz如何規(guī)劃的
- 大師心中的最強功能是什么
- 大師堅定選擇Power BI的核心三點原因是什么
關于Macro可以參考：www.sqlbi.com
以下為翻譯：（水平有限，建議閱讀原文）
Macro Russo的個人Power BI歷程
Marco has been MVP for Data Platform since 2009, from Turin, Italy. He
is a business intelligence consultant and mentor and regularly speaks
at conferences such as TechEd and PASS Summit. Together with Alberto
Ferrari, he founded SQLBI , the world's best resource for tabular data
modeling with PowerPivot , Power BI and SQL Server Analysis Services
Tabular and the DAX language. You can find Marcos personal, BI related
blog at http://sqlblog.com/blogs/marco_russo .
Marco自從2009年開始就是微軟在數據平臺方面的MVP（最有價值專家），他是意大利人，他是BI（商務智能）咨詢顧問以及培訓師，并且在很多相關的重要會議（TechEd and PASS Summit）作為主要發(fā)言人。他和 Alberto Ferrari 一起創(chuàng)立了SQLBI，目前SQLBI是世界范圍內關于Power Pivot，Power BI，SQL Server Analysis Services Tabular以及DAX語言最好的資源。
你可以在這里找到Marco的個人博客：http://sqlblog.com/blogs/marco_russo
（對話開始）
Hello Marco, I am very happy, that my project is constantly growing.
Thanks a lot for participating. The last years of using Power BI
convinced me, this is an amazing product, worth any effort learning
it. My idea with this post is, to share your story as a Business
Intelligence expert, who became a Power BI expert and how it changed
your professional life. Let’s go
Marco，你好，
我很高興我們的項目在穩(wěn)定增長，感謝您的支持。最近幾年使用Power BI越來越讓我認識到這個產品有多牛，再怎么努力學都不為過。今天咱們的主題就是請Macro您來分享下您個人的故事，關于BI方面，您是如何成為Power BI專家以及Power BI給您帶來了怎樣的改變。我們開始吧。
Lars: Can you remember, when you first heard about Power BI/ Power
Excel Add-Ins? How did you get in contact with Power BI?
Lars：您還記得您第一次是什么時候接觸Power BI或Excel中有關Power系列的插件的嗎？
Marco: I used the first version of Power Pivot in August 2009. It was
called Gemini at that time, and I didn’t realize immediately what
would have been its impact in the long term. But at PASS Summit in
Seattle the same year, talking with people of the development team, I
realized that it was something big.
Marco：我在2009年8月就開始使用Power Pivot的第一個版本，那時Power Pivot叫 Gemini，當時誰能想到這個東西會有怎樣的長期影響。但同年在西雅圖舉辦的PASS大會上，我與開發(fā)組的成員交流后，我意識到：要出大事了。
I heard about Power BI in 2014, it was the natural evolution of Power
Pivot, but my concern was the speed of release. When they told “we
will release a new version every month” I said, “I will believe it
only after one year you will do that every month”. I was skeptical,
but it worked well!
我在2014年才聽說到Power BI，它其實是Power Pivot的一種自然進化，而我當時關心的是Power BI更新的速度。當時他們說“我們將每月更新一次”，我說，“好，那要等一年看看是不是能做到”，而結果是，確實如此，真的做到了。
Lars: DAX looks easy at first sight, especially because the syntax is
close to the Excel language. But the whole evaluation context thing is
a lot harder to understand. Additionally, before you write your first
line of DAX, you have to model your data. Do you think it is
realistic, that many analysts, who are now doing their daily work with
Excel, will create their own data models and write DAX measures in
near future? Or in other words: Do you think the term Self-Service BI
is mainly marketing?
Lars：DAX第一眼看上去很簡單，尤其是它的語法和Excel函數很接近。但是計算上下文概念很難理解。而且，在編寫DAX之前，用戶必須做好數據建模。您認為這現實嗎？現狀是很多分析師每天都在使用Excel來處理數據，他們會轉向DAX嗎？或者更簡單說，您認為自助式BI有市場嗎？
Marco: I would separate the answers. If you asked in 1984 whether it
was realistic that every person in accounting would use a spreadsheet,
you would have heard a big “no”. Today we take for granted that on a
mobile phone we can view and edit a spreadsheet. Different users have
different degrees of knowledge of the formula language required, but
anyone can create simple formulas.
Marco：這要分開回答。如果實在1984年所有人都才開始使用電子表格的時候，問這是否會現實，那一定是NO。而現在所有人都開始使用手機甚至在手機里就可以編輯電子表格。不同的用戶對使用公式函數式語言的能力有所不同，但任何人至少都可以寫一些公式。
In certain job roles, a deep knowledge of Excel features is a
requirement. Now, when a table is not enough, what are the
alternatives? Creating a semantic model is a powerful tool, but it
requires skills that we still think are for DBAs or BI Developers
(e.g. data modeling). But many users can learn what they need to
manage multiple tables, and they can refine that over time.
在某些特定崗位是需要對Excel的功能有深度理解的。而如今當電子表格不能滿足需求的時候，誰能任何？構建模型需要功能強大的工具，但人們仍然停留在這些需要DBA（數據庫管理員）或者是BI開發(fā)人員的事情的認識。實際上很多人可以學會直接管理多個表格而且可以直接做到。
Do users who write Excel macros consider themselves “developer”? No.
But they write code. Thus, Power BI users create data models. If you
need to analyze data, in 10 years data modeling will be a required
skill, at least for the basics.
是不是寫了Excel的宏就可以說是開發(fā)者了？不是的。但他們確實也是寫的代碼。因此，Power BI用戶創(chuàng)建模型也一樣（這不屬于開發(fā)）。如果你需要在未來10年中涉及分析數據，那數據建模將是必備能力，至少需要會一些基礎的。
What about self-service BI? Call it marketing, but we always need a
“vision”, a roadmap. I don’t think that 100% of users consuming data
will create models just because they have a self-service BI tool. But
it’s a matter of fact that many users create their own reports, a
smaller number of users customize calculations, and a niche of them
create semantic models. They are all using “self-service BI” or
whatever other name you use, at different levels. Just like there are
many different types of Excel users.
那對于自助式BI有沒有市場，首先我們應該有個愿景以及達成愿景的路線。我并不覺得所有需要使用數據的用戶都需要創(chuàng)建數據模型，即使他有自助BI工具；但他們其中有那么一部分人開始獲利，他們已經開始使用自定義的計算，構建數據模型，創(chuàng)建報告。其實他們已經在不知不覺中或多或少地使用了自助BI（不管叫什么名字都好，只是個稱呼）。這和Excel用戶類似，Excel從初級到精通也有不同級別的人。
Lars: Quick measures were introduced in Power BI Desktop in April 2017
and aim to simplify the creation of measures. Do you think Quick
Measures are a proper way to overcome the complexity of the DAX
language?
Lars：Power BI Desktop在2017年4月的更新中推出了快速度量值以便于簡化度量值的創(chuàng)建。您認為人們可以使用快速度量值而不繞過復雜的DAX語言嗎？
Marco: I have conflicting feelings here. I understand the purpose of
Quick Measures. I don’t like its current implementation. There are too
many limitations, the workarounds could break the calculations (e.g.
enabling auto Date/Time when you have a custom Date table just to use
Quick Measures over dates).
Macro：這個問題是比較糾結的。我知道快速度量值的目的，但只是不喜歡目前的實現方式。因為這里面有很多限制，想要突破這些限制又可能會破壞快速度量值本身（例如：你必須使用系統默認自動生成的日期表）。
It is a feature that could help the occasional user, but they are not
enough for more advanced ones. I know that Microsoft is working on
that, but I would prefer a completely different approach. A DAX
measure is simpler if the data model is well designed. Trying to work
around a bad data model just increase the complexity of the DAX code
required.
這個功能目前可以起作用的場合不多，尤其對高級用戶來說是不夠的。據我了解微軟正在改進，但我希望是用完全不同的方式。如果數據模型設計得好，用DAX寫出度量值也會更容易，相反對一個不好的數據模型自然也就增加了寫出DAX的難度。
A much better solution would be an interaction between data model and
DAX measure that allow to consider both parts of the problem to
wizards/templates defined to solve the very same problems that the
Quick Measures try to solve. Enabling user to write common
calculations in an easy and safe way. Thus, you can look at Quick
Measures, but they are not increasing productivity of advanced users,
and don’t solve all the problems to newbies. There is a lot of work to
do in that area.
一個更好的方式是在數據模型和度量值之間找到平衡點，例如讓快速度量值可以處理很典型的可以做成模板化的模式問題，讓用戶可以用更簡單和安全的方式寫出通用的計算。而現在的快速度量值對于高級用戶既沒有起到作用（例如：他們會用自定義日期表，而不用默認的），對初級用戶也解決不了真實問題，有些雞肋。
Lars: Alberto and you have many offers for people willing to learn
tabular DataModeling and DAX at SQLBI. From books, to in-persona
trainings and video courses. Can you tell the reader something about
that? From newbie to professional: With which of your products should
I start and end?
Lars：在您的網站SQLBI，您和Alberto一起為想要學習表模式建模以及DAX的人提供了很多幫助，從書到培訓到視頻課程，對此您能給初學者一些意見嗎，如果他們需要學習，可以怎樣開始，學習您的什么課程？
Marco: Our mission is to transfer knowledge and accelerate the
adoption of a new technology. We do that providing books, video
courses, classroom courses, conferences, and consulting. We offer
these services on our website, www.sqlbi.com, which hosts our
production of free articles, whitepapers, and tools. We also have a
free video course, Introduction to DAX, which is our suggestion to
start learning DAX. The road to learn DAX also require some data
modeling experience. We have a dedicated web page that suggest the
learning path depending on the products used. We will update and
improve that page in the new website that will be ready before the end
of 2017.
Marco：我們想要做的就是來傳達這些知識，讓人們更快得接收這些。我們提供書籍，線上視頻課程，線下培訓，沙龍以及咨詢服務。我們在官方www.sqlbi.com列出了這些內容，同時也發(fā)布一些文章、白皮書以及小工具。其實我們有一個免費的視頻課：《DAX入門》，如果剛剛開始學習DAX，我建議可以先看這個。而學好DAX其實是需要一些數據建模的經驗的，根據不同的背景，我們給出了相應的一個學習路線圖（https://www.sqlbi.com/learning-dax/）。我們計劃在2017年底再進行一次改版并發(fā)布到我們的新網站。
Lars: From your professional experience as a trainer, what is the most
common mistake, students make when learning DAX? Can you tell us how
to avoid that?
Lars：能不能談談初學者在學習DAX時候，最容易犯什么錯誤以及怎樣避免呢？
Marco: The most common mistake is trying to learn by example. The only
complexity of the DAX is that there are a few abstract concepts that
are unique to this language and do not exists in any other one.
Marco：最常見的錯誤莫過于想通過實例來學習。DAX最大的難點在于其抽象性，這點是和任何編程語言都不同的。
The notion of filter context and context transition is a brilliant
solution to a complex problem. Basically, it is simple to explain and
to learn, but it takes time to digest because it is different from
everything else you already know. I have seen very smart developers
and seasoned BI developers doing the same mistake.
在復雜問題的處理上，理解篩選上下文以及上下文轉換能做出很好的解決方法。但是你會返現，學習這些概念是不難的，但要真正做到深刻理解卻需要時間，因為這和你以前學的都不同。在這方面即使是非常聰明的開發(fā)人員或者BI專家也會經常出錯。
You will save a lot of time if you accept that there could be a new
concept that you have never seen before. I have seen a similar issue
with the introduction of object oriented programming. Some developers
who wanted to explain a new concept using the old syntax were in
trouble. There was something “hidden” that was not working as they
expect. Once you accept that there could be something new to learn,
you are good to go.
學習DAX的捷徑就是直接接受它，不要試圖通過你已經學過的來猜測新的知識。在這方面我在學習面向對象編程時也有過這種經歷。一旦你習慣了用新的模式來思考問題，你就真的上路了。
Lars: In my experience many customers are refusing the cloud as the
place for their business data. At least in Germany that’s a big issue.
But sharing data with Power BI on premise, requires Power BI Report
Server, which is pretty expensive. How do your clients usually deal
with the cloud topic and how do you handle that?
Lars：很多客戶其實不希望把他們的數據存在云上。至少在德國是這樣的。但是如果采用本地部署方案，就需要Power BI Report Server，這個好貴啊。您的客戶是怎樣處理的以及您會怎么做？
Marco: I am in a biased position. My customers call us because they
already adopted Analysis Services or Power BI. Usually, I do not
participate to the presales discussion involving choices between
on-premises and cloud options. I know that there are companies that
want to keep data on premises, and companies that want to go to the
cloud.
Marco：在這方面我并不傾向哪一種。我們客戶基本是已經在使用分析服務或Power BI的人。所以他們其實已經要么在本地，要么在云端，我們基本不參與影響客戶如何選擇本地或云方案的過程。
My point of view about this is that in the long run, all the companies
will go to the cloud, with small exceptions: military and certain
government agencies. These exceptions will probably run the same cloud
technologies on-premises in a controlled and isolated environment.
Outside these exceptions, the question is not if but when. The reason
is a good one: money. Managing the same service at scale in the cloud
is less expensive, once you compute the total cost of ownership. The
time for the cloud adoption will vary based on differences in markets,
culture, policy, laws, bandwidth, skills, and probably much else. But
this will happen. For the same reason why companies no longer develop
their own operating system, their own compiler, their own ERP, and so
on.
至于我個人的觀點是看長期發(fā)展，除了少數例外情況（軍事機構，政府之類），所有企業(yè)都會是在云端的。即使是本地部署也是一種本地私有云。除了例外，其余只是時間問題，核心問題是：成本（費用）。云端解決方案將降低總擁有成本。而這個趨勢取決于市場、文化、政策、法律、品牌傳播、人員技術等，但遲早會這樣。正如為什么沒人再自己去開發(fā)操作系統或ERP也是一樣的道理。
We can learn from the short history of computer science that what
seemed impossible to change in a decade, was completely outdated a few
years later. I see that technology adoption in Europe has a much more
conservative approach than in other continents. But I do not see a
single technology that has been adopted in United States but not in
Europe (maybe years later). Well, Europe skipped cable TV going
straight to satellite TV. But it is probably the exception that proves
the rule.
從最近十年計算機技術的發(fā)展來看，歐洲相對來說是比較保守的。但我還沒發(fā)現有什么是在美國適用而在歐洲不適用的案例，歐洲隨后也會一樣。當然，歐洲沒有經歷有線電視直接進入衛(wèi)星電視應該算是一個例外。
Lars: With SQLBI you founded OKViz some time ago and provide your own
custom visualizations. How critical to Power BI’s success is the fact
that the visualizations are open source and can be developed from
everybody who is technically capable?
Lars：我們發(fā)現除了SQLBI，你們還成立了OKViz，專門提供自定義可視化。Power BI自定義可視化是開源的，這是Power BI成功的重要因素，只要懂相關技術就可以做嗎？
Marco: With OKViz, we created a brand for custom visualizations. As of
today, it is a “pure startup”, because it has costs without producing
any revenue. The marketplace is not ready, the platform is not ready,
and custom visuals are second class citizen by now. Hopefully this
will change, and OKViz will be able to increase the production of
custom visuals once there will be the conditions for a sustainable
economic model.
Marco：對于OKViz，我們發(fā)布了一系列自定義可視化。直到現在仍然是起步階段，因為目前這塊沒有任何收益。市場還沒有完全具備，平臺也沒有完全理想，自定義可視化目前仍屬于二等公民。我們希望這種狀況可以改變，只要有適度的收益，OKViz將不斷增加產品。
At the moment, custom visuals are an important tool for Power BI
success, but in terms of real adoption, native visualizations are much
more used. The biggest challenges are versioning, support, and
automatic updates. Creating a stable platform for custom visuals in an
environment that evolves every month is incredibly hard. We can easily
see the bugs, but not the amount of hidden work to sustain that.
就目前來看，自定義可視化對Power BI的成功來說確實是重要的工具，但在實際使用方面，內置的可視化用得更多，最大的問題就是版本升級以及自動更新。來應對每個月的升級提供相應的自定義可視化產品升級很難，很容易就會出現問題，在這背后需要大量的工作。
I am optimist for custom visuals in Power BI if this will create a
true marketplace. I am skeptical about enabling anyone to create
custom visuals. We will never see Power BI users writing
visualizations just as an Excel user writes a macro. What is possible
is enabling an existing developer to create custom visuals for Power
BI. The open source is a very relative advantage. The entire platform
is not open source, you cannot extend the code of the platform, you
can just copy and adapt other custom visuals. Native visuals are
written without using the custom visuals platform, so you cannot
derive custom visuals from native visuals.
如果在Power BI的自定義可視化領域確實有市場，我們會加大力度。我覺得不可能讓任何人都能創(chuàng)建自定義可視化。我們不會讓Power BI的用戶去寫自定義可視化就像沒指望Excel用戶去寫宏。可能一部分開發(fā)人員會進來做有關Power BI的自定義可視化。自定義可視化的開源在一定意義上是有優(yōu)勢的，如果平臺不開源，那就無法擴展平臺的能力，你只能使用已有的東西。而目前默認的內置可視化并沒有采用開源的策略，導致我們不能從內置可視化來擴展，只能完全重新構建。
To recap: I believe custom visuals are extremely important, and we
invested a lot on them. However, I would not sell custom visual as
something that any Power BI user can adapt according to any specific
requirement. Learning curve to write custom visuals is not so quick.
總結的話，我認為自定義可視化是很重要的，我們已經在這上面投入很多。但目前沒有在這上面產生收益。學習自定義可視化的開發(fā)并不容易。
Lars: If you could ask for a new feature in Power BI (what everyone
can via ideas.powerbi.com) what would that be?
Lars：如果讓您在Power BI社區(qū)提一個最希望的新功能，您會提什么？
Marco: My number one is opening API to interact with Power BI Desktop.
Let me ask the vote for this:
https://ideas.powerbi.com/forums/265200-power-bi-ideas/suggestions/7345565-power-bi-designer-api
Marco：我最希望的是Power BI Desktop可以開放API。我已經在這里投票：https://ideas.powerbi.com/forums/265200-power-bi-ideas/suggestions/7345565-power-bi-designer-api
This feature would enable the creation of many new tools and
solutions.
如果可以實現這個，那么將有更多想象空間，可以做很多新工具和解決方案。
Lars: Could you name 3 things about Power BI, which make it a great
product?
Lars：如果說Power BI是很棒的產品，只能選三個優(yōu)點，您會選什么？
Marco: ? Free access to almost all of the features to start a proof
of concept. ? Integration between Query Editor (M), data model (DAX)
and visualization. ? Computing power of the engine: it is very hard
to find something faster on the market.
Marco：
1、可以免費使用幾乎所有功能。
2、將查詢編輯（M），數據建模（DAX）以及可視化完美整合。
3、強大的計算引擎：在目前來說，很難找到比它更快的了。
Marco, thank you, Alberto, Daniele and all your colleagues behind the
scenes, for all the great posts, books, products and community work. I
know, there are many people out there (including myself) who are very
thankful for all the work you do. To get in touch with Marco, please
use the following website.
好的，非常感謝Marco，Alberto，Daniele以及在幕后的SQLBI的伙伴，感謝你們帶來了這么多好的文章、書籍，產品以及公益活動。我相信，和我一樣有很多人也是如此。如果大家希望聯系Marco，請登錄如下網址：http://www.sqlbi.com/author/marco-russo/
完。（翻譯 by bi@excel120）
總結
大師之所以是大師，從09年開始已經8年過去了，還記得嗎：
在你認為值得的事情上花10000小時，你就會成為大師。
不知道您聽了Macro的這些內容有何感想，至少我有以下幾點感觸：
- Power BI是一個幾經考量而進化來的產品，并有強大社區(qū)支持，將繼續(xù)進化。
- 近十年時間說明大數據仍處于概念階段，而商務智能，尤其是人人可用的自助式商務智能會先于大數據登門數據時代舞臺，成為重要角色。
- 初學者如何快速學好Power BI的本質在于駕馭DAX的抽象性以及用空杯心理接受DAX的全新概念，切勿猜測或帶有過往經驗判斷。
- SQLBI.com提供了一套免費的《DAX入門》視頻教程，大家可以在www.sqlbi.com的在線培訓中免費獲取。當然，該視頻只能在線播放，但配備了英文字幕。
- Power BI（Desktop）有很大優(yōu)化和想象空間，微軟的數據工具不是工具，而是一個全新的大生態(tài)，我們只不過是在這個數據生態(tài)中生活工作而已。
Power BI是現代自助式商務智能的落地體現，掌握Power BI及數據建模，將幫助我們在數據時代降維攻擊停留在原來時代的人和企業(yè)。
技術不停，學習不止。
