Microsoft's new PowerPivot from a QlikView standpoint
Posted on Fri, Jan 22, 2010
I recently had the pleasure of attending a Microsoft Solutions for Business Intelligence seminar in Dublin. The seminar will travel to 39 countries over the next three months. Dublin was the third stop on this tour. The presenter was Rafal Lukewiecki who works for a company called Project Botticelli. Rafal is a very polished presenter. He has a natural flow and covers every detail in both his dialog and his accompanying demos. As a presenter he is very impressive.
A large portion of the day was focused around Microsoft's new "In-Memory" B.I component PowerPivot. PowerPivot enhances the B.I capabilities of Microsoft Excel and was on more than one occasion referred to as "Excel on Steroids". It is being marketed as "Self Service BI" which has the potential to empower the end user to find the answer for themselves provided they can access the appropriate data sources. This sounds good in principle but without the content management facility provided by SharePoint it could intensify the silos of information that currently exist with Microsoft Excel leading to more versions of the truth.
My initial reaction to the tool is that I am quite familiar with many facets of it from my Qlikview and Qliktech experience. It is testament to Qliktech and their growth that Microsoft should adopt such a similar methods. Some of the similarities include.
PowerPivot uses an extremely similar approach to Qlikview in relation to how PowerPivot works under the bonnet. It uses patented compression technology VertiPaq to optimise the amount of disk space taken up by a PowerPivot "Application"(the compression rate is about 10:1) and stores data as a numeric value which has a pointer to the actual value. It is uncanny how similar this is to Qlikview's patented in memory technology. "Application" is the common term for a PowerPivot file at present but may change. The capacity for huge volumes of data was used as a wow factor throughout the day but from a Qlikview perspective there was nothing that has not been done before.
Sample applications are available from the PowerPivot website. They have an all too familiar look and feel and in keeping with the proven Qlikview model provide the facility to explore the product features using information that is meaningful to the end-user e.g. Sporting information, aviation information, nutritional information.
One of the wow factors of the demo was a component in PowerPivot called a slicer. A slicer is the PowerPivot equivalent of a Qlikview listbox and serves as an "on the fly" filtering mechanism. Any previous exposure to Qlikview will negate the excitement provided by this feature.
It is rumoured but has not yet been confirmed that PowerPivot will be cost free for users. There is the cost of SharePoint 2010 if users wish to share PowerPivot applications using in a centralised content management space. This is similar to the Qlikview Personal edition model adopted in Qlikview 9 albeit not identical. The idea seems to be that increases in user adoption will feed user consumption which in turn increases revenue. Perhaps this marketing method is a way of encouraging a user to at least consider PowerPivot/Qlikview when a user is leaning towards an OpenSource B.I. tool such as Pentaho, JasperSoft, or SpagoBI.
Other issues I would like to comment on are listed below.
- DAX(Data Analysis Expressions)
DAX is the new language for expressions in PowerPivot and is quite a powerful language. In empowers the end user to do multidimensional analysis and has a syntax similar to Excel Functions. It has a predefined list of common functions that can be performed on measures, strings etc. This is a strong feature for PowerPivot as relative time comparisons can be performed with ease. This is in contrast to my limited exposure to IBM Cognos' In-Memory B.I. tool TM1 which is part of the Cognos Express suite. I attended a Cognos Express training course in September 2009 and relative time comparisons were an obvious weakness during the course.
The Microsoft Suite of BI products have traditionally consisted of too many components without the natural fit of some of the larger BI vendors i.e. Cognos, Business Objects. The Microsoft BI subcomponents of BI have changed with the release of SQL Server 200 R2 and is now composed of three layers, Microsoft Office 2010, Microsoft SharePoint 2010 and SQL Server 2008 R2. This model will flatten further into the future as the subcomponents become more entwined. This move makes sense but there are still too many individual parts within each of these layers. The SQL Server component consists of the SQL Server Engine, SSAS, SSIS, SSRS, and the new Microsoft Master Data Services. The SharePoint 2010 layer is composed of Performance Point Services(the artist formerly known as Performance Point Server). Performance Point Services incorporates ProClarity. On a positive note the Decomposition Tree is one of the available chart types and has been inherited from the afore mentioned ProClarity. Visually this is a really nice chart in my opinion and I have always been a fan. I would like to see something similar in Qlikview in the future. Perhaps the next version of Qlikview will provide this. Another positive was some of the Silverlight charts shown in the Performance Point Services demos. These charts were very pleasing on the eye. As is typical with Rafal Lukewiecki the demos were flawless. I have to commend him on his presentation skills and his ability to communicate.
Microsoft Office 2010 is a requirement for PowerPivot. Larger organisation are traditionally slow moving beasts when it comes to upgrading to the latest version of office. For this reason the exposure of employees of these large organisations may be limited to non-working time. Microsoft have a huge market however and any significant portion of this market is a huge customer base. Positive initial feedback has the potential to influence adoption rates and tap into these numbers.
In summary I am looking forward to getting my hands dirty and testing the functionality of PowerPivot first hand. I am sure given the time constraints involved the features shown in the demos only scratched the surface of the capabilities of PowerPivot. It strikes me that I will appear to have a biased opinion when reading this blog. This is not my intention and I am open to being convinced by the performance of the tool. I have low expectations at this juncture. At present PowerPivot feels like an attempt to play catch up to Qlikview. I do however feel that interesting times lie ahead as mass adoption will lead to further improvements and perhaps a more even contest some time in the future.
Darren Kerfoot - Senior Analyst with QlikPower