BeyeBLOGS | BeyeBLOGS Home | Get Your Own Blog

« My 10 favourite business intelligence blog posts of 2009 | Main

January 29, 2010

QlikView from a PowerPivot standpoint


A week or so Darren Kerfoot of QlikPower, a QlikView consultancy, wrote a thought-provoking blog about "PowerPivot from a QlikView standpoint." Please do read it: I found it nicely balanced. The thought it provoked in me, and which I tweeted, half joking, was that I might blog from the other side of the fence. I was quite surprised by the number who said I should. So here goes ... what does QlikView look like from the PowerPivot standpoint?

I will follow the same headings as Darren uses, to make comparisons more readily, although I should say that perfectly I would structure a full comparison somewhat differently.

I will be quite critical of some aspects of QlikView, but let's be clear from the start. In general I think QlikTech are a very smart company. They have excellent growth and great customer satisfaction. Personally, I think that mostly comes from compelling marketing, an innovative and effective sales process, and excellent customer support; the product is good enough to sustain this. QlikTech are an excellent Microsoft partner and it's good to see their success.

Darren does say that his initial reaction to PowerPivot (on seeing it demonstrated by the ever-admirable Rafal Lukawiecki) was that it was quite familiar to him from his QlikView experience. Our hope, as the PowerPivot product team, is that it will be even more familiar to Excel and SharePoint users. In fact, I might say that the innovative features of PowerPivot are less important than what we haven't invented: there's a great advantage to our Excel-like querying interfaces, our Excel-like syntax for expressions, our use of Pivot Tables for analysis, and our use of the familiar SharePoint document model for publishing and managing analyses, and so on.

Underlying Technology
As Darren says, PowerPivot, under the bonnet, uses an in-memory store. QlikView does too. Then again, so does Tibco Spotfire, IBM TM1, Advizor, PivotLink and Altosoft to name only a few in the BI space. And of course column-stores, such as the Vertipaq technology that PowerPivot uses, are common in the relational database world, while Sybase, Oracle and IBM also have in-memory relational. As QlikView does not use a column store, PowerPivot is really more like those other systems, with QlikView being the exception. The differences are minor for the end user: QlikView appears to compress strings more effectively than PowerPivot, although that can catch you by surprise when the compressed data is uncompressed back into memory; PowerPivot appears to perform much better with calculated columns. I haven't done anything like benchmarking on this - just my own observations from running both on the same box.

Talking of engines, some have been misled to believe that QlikView's supposed "associative analysis" represents some significant engine smarts. I have even heard analysts very misleadingly say that QlikView has "association rules" - implying some kind of data mining, such as Microsoft implements in its Data Mining server and Excel Add-ins. QlikView add to the confusion by talking about an associative "architectural model." However, despite the hype, as Curt Monash points out (or rather, painfully extracted from QlikTech themselves through a long thread of comments) it is not so: "The associative aspect is really more meaningful in describing the end user experience, in that you see visually what is associated and is not associated with any particular selection or drilldown." As Curt says, "Thank you for admitting that clearly!!! It wastes a fair amount of analysts' time when your company pretends otherwise." http://www.dbms2.com/2008/08/04/qliktech-qlikview-update/

So I guess we all look pretty similar. Of course, at Microsoft we think we have some particular smarts in our engine, but in general the data-handling capabilities should be similar. I do hear of QlikView customers having difficulty scaling - it will be interesting to find PowerPivot's limits, too. It's early days for that, of course, but I'm sure we'll find them.

Sample Applications
Darren is right, our sample apps on www.powerpivot.com are pretty familiar. There are a limited number of public data sets out there to share: sports, and so on, are common ground. So many of these sample application sites are similar. Out of context for this post, but still well worth a visit, is the sample page of our friends at Tableau. http://www.tableausoftware.com/learning/examples Now folks, THAT's a set of sample apps!

Slicers
I think Darren's right - QlikView users will see nothing too exciting here. However, our target audience of Excel users (and especially PivotTable users) do like Slicers very much. I think they are pretty basic in the first version - in the future, expect to see an even better experience. Nevertheless, they are a very natural step up from traditional pivot table filters, and provide a nice visual interface for those functions: and they work in Excel services too, bringing browser interaction alive for those users who consume, rather than produce, PowerPivots.

Market Exposure
We have long touted the idea of releasing the PowerPivot add-in as effectively a no-cost download. In fact we were talking about this publicly before QlikTech made their Personal Edition free on the desktop. Maybe they were hoping to pre-empt us? Maybe not. Probably, they just thought it was a good idea.

After all, it is a good idea. For Microsoft, we had already released the Data Mining Add-ins for Excel as a free download, so we knew this model was attractive. For QlikView, it was a good way of heading off complaints about the high cost of ownership. High TCO is still the top complaint I hear about QlikView from their customers, even though their pricing model seems, at first glance, quite modest. However, we see many cases where consulting fees have grown dramatically for QlikView users, who often bought into a story of software that was so easy to use that applications could be built in hours. It can be true in some cases, but there are a helluva lot of QlikView consultants out there, doing very well for themselves, and rather contradicting that myth.

There's a similar story around partnerships. QlikTech have been quick to say "no data warehouse required" and aim to reduce the dependency on the IT department: music to the ears of many business users. Yet you only have to look to QlikView's partnerships to see the weakness in that argument. Some examples:

:: QlikTech Announces Support for HP Neoview: http://www.qlikview.com/Contents.aspx?id=8702 "Robust data warehouse provides a scalable foundation on which QlikView delivers user-driven analysis"
:: QlikTech and Informatica: http://www.qlikview.com/Contents.aspx?id=9270 "The combination of Informatica's robust data integration products ... with QlikView ... enables enterprises to optimize their entire data management process"
:: QlikTech and Kalido: http://www.qlikview.com/Contents.aspx?id=6628 "For QlikTech customers, Kalido provides a robust, enterprise-ready information management capability."

Perhaps QlikTech's marketing team just need a new thesaurus, but it sure looks like there's a problem in the delivery of "robust" solutions that requires data warehouse and data integration partners to solve.

This, ultimately, is the big difference between the market exposure of PowerPivot and the market exposure of QlikView. PowerPivot is one part of a seamless story stretching from robust and scalable enterprise data solutions, such as our Parallel Data Warehouse, through middle-tier applications such as SQL Server Integration Services, down to flexible personal analytics on the desktop with Excel. QlikView, great application though it is, offers only one part of that story: on the one hand it must pull in partnerships for robust enterprise applications, and on the other ... well, even QlikView needs "Export to Excel" to go the last mile in flexibility and agility.

DAX (Data Analysis Expressions)
DAX is a winner. Users can start from simple Excel-like expressions and build-up to really quite sophisticated dimension-navigating, time-aware functions that return in-memory table objects for further, nested, functionality. I wish it was easier at the advanced level, but so far users are really delivering great applications with DAX. There is a contrast with QlikView that Darren does not call out. QlikView requires a LOT of scripting. When QlikTech presented at the Boulder BI Brains Trust last year, this was noted in several tweets during the demo - you really need to know your way around Visual Basic scripts to get the most out of QlikView. This is certainly a problem. I know, because we used a lot of scripting in a previous product on which I worked: SQL Server Integration Services. In fact, I wrote a book just about the use of script in that application. I can tell you - scripting is not for business users.

There is an irony here. As a developer, in some cases, I might actually prefer a script to some of the complex nested expressions that I build in DAX, even with the auto-complete and parentheses-parsing tools we provide. But for business users, that certainly isn't the case. The flow of control needed in scripting is just not how they think: and the advanced Excel user in marketing or finance, is often quite expert at building and debugging complex functions.

Summary
I hope you have found this interesting. Let me re-iterate. QlikView is an excellent application, and QlikTech sell it in a very compelling manner. I don't expect PowerPivot will put a huge dent in that, because it really is the marketing and sales process, rather than the software, that have brought QlikTech their remarkable growth. QlikTech are especially good in small geographies, where their hard-driving, high-octane, hands-on, move-so-fast-they-can't-see-the-problems style works very effectively, especially when they reach business users directly, rather than IT. In the larger, more mature, and better-served US market they still grow strongly, but don't have the same mindshare.

I do expect that PowerPivot will lengthen sales cycles for the QlikView team in many cases. However, I also expect that in many organizations we will co-exist quite happily. Excel power-users will love and use PowerPivot. Users who enjoy QlikView's polished UI and navigation tools, will no doubt still enjoy that experience. These do appear to be two separate groups of users. I find very few Excel power users to be QlikView converts.

Darren is certainly right: interesting times lie ahead!

For more information on QlikView, see http://www.qlikview.com

For more information on PowerPivot, see http://www.powerpivot.com

Posted by Donald Farmer at January 29, 2010 5:30 PM

Comments

I will be the first to admit I bleed blue, Microsoft technology is 99.9% of what I use every living day as a BI professional...

With that statement out of the way I would value the opportunity to add my 2, perhaps 1.5 cents.

Microsoft' strategy, marketing, and placement of PowerPivot represents what I would call in laymen's terms the 'truth' for where and when the product should be used. The 'truth' being for ad-hoc, self-service analysis needs and wants that IT simply cannot nor perhaps should not provide. It is as 'the man' here says "one part of a much larger MSFT BI story".

Some other software companies have not marketed their self-service products entirely truthful and at times have even won customers that truly needed Corporate BI. Yes, Microsoft did come to the self-service party a tad late but atleast the entered the market on morale grounds.

Maybe I'm just on overload in regards to Corporate Vs Personal BI models because of the sheer volume of inquires I have been increasingly received or maybe software companies should focus more on helping their customers business grow then their internal sales goals.

Derek Comingore
Microsoft MVP

Posted by: Derek Comingore at January 29, 2010 6:32 PM

The misunderstanding on the associative aspect of QlikView is a communication problem. On the side of QlikView, the associative aspect is merely part of the user interface. We've long understood that and end-users also pretty much get it.

As to the partnerships, I don't think it's a sign of weakness. Being partners means growing together and building on each other's strengths. It does not only mean that QlikTech needs partners to deliver end-to-end solutions, but also that QlikTech's partners need QlikView to enhance their overall offerings.

Posted by: Kevin Yapjoco aka @analyticbits at January 29, 2010 7:21 PM

The data warehouse question is a common misconception about Qlikview. Personally I'd rather build Qlikviews with a solid data warehouse as the data source a sound conformed dimensional data model removes much of the complexity that must be overcome when integrating data from multiple systems. However, if you don't have a data warehouse you don't have to wait six months to start using Qlikview while you build your DW. In addition with proper layering of Qlikviews you can achieve a pseudo-DW completely within the Qlikview platform.

Posted by: Chris Cammers at February 4, 2010 7:13 AM

It's always an interesting partner cohabitation story with Microsoft and tech partners.

QlikTech is "an excellent Microsoft partner" as Donald puts it, but they are going to face-to-face challenge them with PowerPivot and have to compete in the same market space.

The success of Microsoft is mixed in areas where they challenge niche areas that have established players and are not broad-based technology frameworks. (i.e. PPS)

Posted by: Mark Kromer at February 7, 2010 12:58 AM

Dear Mr. Farmer,

My name is Ella and I blog about BI in the Israeli Microsoft platform:
http://blogs.microsoft.co.il/blogs/barbaro
I understood you’re coming to Israel to lecture at the coming SQL & BI Data Platform event, (May 30th and 31st).
Any chance I may ask for 5 minutes of your time for a short interview?
I would REALLY REALLY appreciate it!

With thanks in advance,
Ella

Posted by: Ella Maschiach at May 4, 2010 1:57 AM

Hi Mr. Farmer,

How are you?

When we last spoke, you said you could give me an interview for my blog at Microsoft Israel.
I know you'll be in Israel the coming Sunday and I wanted to ask – can I have 24.59 minutes of your time for your interview? Coffee on me! :)

Any chance you have time Sunday afternoon, after your seminar?
I would really like to sit with you somewhere quiet and bring a cameraman with me.
Could I possibly ask you to get back at me on Facebook or through the email?

With thanks in advance,
Ella

Posted by: Ella Maschiach at May 25, 2010 2:01 PM

Hi Donald,
I agreee with you on almost you wrote.

Some points:
- Data Mining in excel (including Excel Add-In) is near the end, correct? The creation of Predixion company maybe is a confirmation of that.

- I believe in Power Pivot only if teh source is a data warehouse (OLTP) or a OLAP database. If not data quality problems naturally arise.

Best regards,
Pedro Perfeito
Microsoft MVP
TDWI - CBIP
www.pedrocgd.blogspot.com

Posted by: PedroCGd at July 13, 2010 5:52 AM

Nice to see a comparison of these two tool. I have used the MS BI stack up to and including the PerformancePoint Server release but not PowerPivot. I have used Qlikview for the last year or so.

Some very important points for me.

1) I read PowerPivots cannot be > 2 gigs ( http://msdn.microsoft.com/en-us/library/ff628113.aspx ). Not sure how well the data compresses but this could be a deal killer for big organizations.

2) PowerPivots require sharepoint for update data from data sources and for serving up. This is fine i guess but it would be great if secuirty and data updates could be easily mangeed outside of sharepoint so that a thick client deployment without sharepoint was possbible.

Posted by: Tim Webber at July 28, 2010 3:05 PM

it's very useful post thanks for sharing it, and what about silica.

Posted by: Tas Flowrance at September 20, 2010 3:27 AM

Donald - I have ue BO, Cognos, OBIEE, SSAS, and Tableau. I also created a decent sized Qlikview applicaton. What impressed me & the users of the Qlikview application, other than the rocket speed of development, was the insight into the data which in which they were NOT exploring. I.E. the data that did not match their exploratory filtering. Qlikview does not filter "unwanted" data by making it disapear - that data simply turns gray - you still see it, so can gain useful insight on the data that did not meet your filtering. HUGE benefit (read that again). No other product on the market does that.

Posted by: Dave at September 20, 2010 8:41 PM

Post a comment




Remember Me?