Why is Vertipaq so darn fast? (video blog post)

I posted a couple videos explaining why the Vertipaq engine approach yields such high performance.  Included is a simplified explanation of the architecture as well as a demo of Vertipaq performance against a reasonably large data set (> 200M rows).

This is a good watch for anyone who's heard of Vertipaq (and even used it) but may not have a detailed understanding of why it's design yields such good performance.

Posted PerformancePoint Bing Maps Customization to my Video Blog site

Recently we've been working on some geospatial customizations.  One of these is a Silverlight customization that incorporates a highly interactive Bing Maps layer into a PerformancePoint dashboard.

I posted a full walk-through demo on my video blog site.

Bing Maps PerformancePoint Customization

Class not registered REGDB_E_CLASSNOTREG when browsing cube in BIDS

It seems like I encounter this problem every few months when using a new server or workstation that doesn't have old 32-bit office installed on it. This posting more or less a quick link for me to re-find the OWC component link, but maybe it'll help others from time to time...

Here's the link:
http://www.microsoft.com/download/en/details.aspx?displaylang=en&id=22276

And here's what the error looks like:

iPad dashboard app for SQL Analysis Services

On a recent project I put together a BI dashboard app for the iPad that uses SQL Server Analysis Services as the back-end data source (demo video below).  

For those who don't know, I used to be a mobile developer 100% of the time.  Of course this was back in the Palm/Pocket PC days when nobody ascribed much value to mobile apps.  I diversified into other areas where customers were easier to find than a needle in a haystack :-)...but have never really lost interest in mobile apps, and still enjoy it.

Anyway, here's the demo video for the mobile dashboard app:

All mobile apps will be HTML5 someday (maybe)

I've been involved in mobile apps since before they were so fashionable.  I totally loved my first mobile device--a Palm Vx PDA (remember that term...PDA?).  It was (and still is) in my opinion the best mobile device ever designed.  I still have it--and no, I don't still use it! 

That Palm Vx inspired me to change careers and develop mobile applications for a few years. One of my favorite was a paint matching app on the Palm platform for a large chemical company.  It put an entire microfiche library for automotive color matching data on a handheld device.  The field reps loved it.  It was one of the most interesting projects I ever worked on--but it wasn't a mainstream use case by any stretch.

Fast forward...a new decade has come.  Mobile devices are now fashion statements, and low-power, long-battery life tablets are going mainstream--replacing PCs for many home and business users.  I always knew it would happen, and it's cool to see it.  

On a recent project, I put together a BI dashboard app for the iPad that uses SQL Server Analysis Services as the back-end data source.   This is the type of thing I usually do using web-based tools.

While I was doing the development, I kept having the feeling that the end-user experience users get from native mobile apps are so much richer than much of the the software we now use that's delivered in a web browser.  Does anyone else feel this way?

Is it possible users are adopting mobile devices so quickly in part because the software is more enjoyable to use than the web-based alternatives?  If so, then will HTML5 really supplant native applications and become the unified theory of mobile app development?  I'm not so sure the death of native mobile apps in favor of HTML5 is one of those inevitable things I believe in...

Installing #PowerPivot 2012 CTP3 for Excel on a Terminal Server

Had a bit of trouble getting the CTP3 version of the PowerPivot add-in working on a Terminal Server today.  This terminal server is a Windows 2008R2 server running remote desktop services, and had the 2008R2 version of PowerPivot installed.

I was able to install it as an administrator, and use it with the administrator account I used to install it.  However, after logging in as a different user, I receved an error after launching Excel:

There was an error during installation. Downloading file:///C:/Program Files/Microsoft Analysis Services/AS Excel Client/10/ Microsoft.AnalsysServcies.Modeler.FieldList.vsto did not succed.

 

The error is basically saying that the old version of the add-in cannot be found, so it seems that doing an in-place upgrade on a Terminal Server is not yet addressed in the CTP cycle.

I don't really want every remote desktop user in the terminal server running the add-in installer themselves (and they won't have priviliges anyway), so that's not a good solution.

To resolve the problem, I tracked down the add-in registration keys in the registry for the admin user account I used to install the add-in, and saved them out as a .reg file. 

If you have the same problem, create a text with the extension .reg, add the following to it, and have PowerPivot users double-click to run this in their terminal session (alternately you could design a script to do this for them).

Contents of upgradepowerpivot.reg:

Windows Registry Editor Version 5.00
[HKEY_CURRENT_USER\Software\Microsoft\Office\ Excel\Addins\Microsoft.AnalysisServices.Modeler.FieldList]
"Description"="Microsoft SQL Server PowerPivot for Microsoft Excel"
"FriendlyName"="PowerPivot for Excel"
"LoadBehavior"=dword:00000003
"Manifest"="C:\\Program Files\\Microsoft Analysis Services\\AS Excel Client\\110\\Microsoft.AnalysisServices.XLHost.Addin.vsto|vstolocal"
"CartridgePath"="C:\\Program Files\\Microsoft Analysis Services\\AS OLEDB\\110\\Cartridges\\"

MSBI - We've come a long way, baby

It's the eve before SQL PASS 2011, and as I look over the conference agenda I can't but reflect on the progress Microsoft has made in establishing itself as a serious BI contender, and how lucky I was to get involved early on.

When I first started working with Microsoft's BI stack, I don't know that I'd call it a "stack" really.  It was more a recognition that there were these various pieces of Microsoft products that--if you took the time to stitch them together with some other third-party products--made a pretty compelling solution.  What's more, it could be done for a fraction of the cost to use what were then the specialized heavy hitters in the BI space.

SQL Server had the RDBMS, SSAS 2000, and some basic ETL capabilities with DTS. It even had everyone's favorite stealth feature: data mining.  Reporting services made a good UI that didn't require additional licensing. For better analytical capabilities ProClarity fit the bill.  Later, Business Scorecard Manager (much later PerformancePoint Monitoring and Analytics) could be considered at the high end.  All of this worked great together, but surely unless you knew about all the pieces and how to engineer them, it just wasn't happening for most end-users.  Consultants (like our firm) knew how to put it together, but you can't reach everyone in the world that way--usually only those who "know a guy who knows a guy who knows the right consultant for something like that"...but not the kind of awareness that drives a mass market strategy.

It's been a cool adventure to watch the successes with MSBI, like the successful integration of PerformancePoint and SSRS with the SharePoint portal; the new and innovative features like Excel Services and BCS and PowerPivot.  It's not been all fun and games, as there have been some failures--PerformancePoint Planning; the acquisition and controversial demise of the ProClarity assets.

The future, though, looks pretty bright.  As field implementation consultants, we used to have to convince prospective clients that the myriad of products we were talking about really did comprise a cohesive strategy (a typical response: "That sounds awfully complicated.  We're going to go with Brand X because their product is 100% integrated (or so we've heard)". We don't have those conversations anymore.  "Microsoft BI" is as natural conversation now, and the messaging is mostly cohesive.  Each year Gartner validates the power in Microsoft's strategy to democratize BI and make it a standard part of an IT architecture rather than a luxury for the well-heeled.  The vision is right, and nothing happens without a good vision.

Even on data volumes and scale we've come so far.  A few years ago parallel DW systems were a far off, large scale thing that was beyond what we as mere mortal MSBI pros would consider in our wheelhouse.  Now we have PDW which draws on the skills we already have mastered--and in some ways simplifies things so some of the skills we have aren't as necessary in the appliance environment. Amazing.

There is still more to accomplish.  The products are much better integrated than before--but there are still some misalignments to polish up.  We still have too many languages and dialects to learn (how do you convert a string to a date?  Well, it depends whether you're using SSIS, SSAS, SQL, MDX, DAX, or SSRS!!).  And we still have to put footnotes on demonstrations ("what I just showed you only works if you use underlying technology A, and since you're using option B, this UI won't work for you unless we change some architecture first").  But success is a journey, and I'll take where we are over where we were five years ago every day of the week and twice on Sundays.

It's just amazing how far we've come together--Microsoft, the partners and the customers.  We lose sight of what a transformation has occurred...at least I do sometimes. If one considers how quickly we've come this far, I can only imagine what kind of solutions we'll be building with MS BI technologies only a few years from now!