Jamie's Junk : Analysis Services

SQL Server 2008 and Analysis Services DMV's

JamieMac — Sat, 22 Sep 2007 01:23:00 GMT

One of the cool new features coming in SQL Server 2008 Analysis Services are Dynamic Management Views (DMV). Actually I never thought of them as DMVs coming from my background, but since everyone else calls them that, there's no reason we shouldn't either.

So what are they and how did they come about?

DMX users are most likely familiar with the syntax:

SELECT * FROM MyDataMiningModel.CONTENT

What you may not know is that this is really a DMX query against the MINING_CONTENT schema rowset - we simply exposed via the SQL-like syntax of DMX for convenience, placing a schema restriction on the model name.

While we were implementing some features around resource monitoring, we realized we could expose the information we provided through this mechanism and make it much more usable than schema queries, since most query tools can't do them.

So what we did was to add the ability to query any arbitrary Analysis Services schema rowset through DMX. For example

SELECT * FROM $SYSTEM.DMSCHEMA_MINING_MODELS

returns the MINING_MODELS schema. As you would expect, you can select any subset of columns, filter by arbitrary where clause, and for those adventurous enough, use as input to other DMX statements - e.g. INSERT INTO or PREDICTION JOIN.

In my opinion, some of the most interesting applications are in integration with other BI tools. For example, you can now create SSIS packages that operate on every dimension, cube, mining model, etc. You can create a selector in Reporting Services populated with the result of an AS schema query. You can even create AS cubes based on AS cube information!

Some examples:

SELECT * FROM $SYSTEM.MDSCHEMA_SETS WHERE SCOPE=2 // return all session named sets

SELECT SERVICE_DISPLAY_NAME FROM
$SYSTEM.DMSCHEMA_MINING_SERVICES
WHERE MSOLAP_SUPPORTS_DATA_MINING_DIMENSIONS // return names of algorithms that can create OLAP dimensions

BIDS Helper available on CodePlex

JamieMac — Wed, 16 May 2007 21:57:00 GMT

Furmanng posted a set of utilities to CodePlex that help your OLAP development called BidsHelper.

The features are directly integrated into BI Development Studio in toolbars and menus inside the Analysis Services designers. It currently has an aggregation manager, calc helpers, MDX Script deployment utility, dimension health utility, plus many more features. Even if you weren't interested in the specific features Furmangg put together, it's worth checking out to see how he was able to integrate at that level.

So, check it out!

Awesome job op on the Analysis Services team

JamieMac — Mon, 05 Feb 2007 11:04:00 GMT

Any readers out there that are awesome UI developers with strong ambition and experience shipping packaged software? It's been a few years since such an opportunity has come up - it fact, long before I became dev manager of the team, but now we are looking for a strong developer to come join our OLAP tools team. This is the team that is responsible for all of our great OLAP UI used to create dimensions and cubes and all the accourements such as custom calcs and translations, aggregations, partition management, and so on. Not limited to that - this team also takes on the Analysis Services API layer, including OLE DB, ADOMD.NET, and AMO (don't forget DSO9 as well!). Oh yeah - they also do all the slick integration with Visual Studio and our management tools - anything I've forgotten? Migration wizard - that too.

There's quite a lot of surface area for you to take on if you join this team. You get to work with an incredible team of UI/client devs, plus the OLAP engine and data mining teams, plus you will be working with other teams around SQL Server and Microsoft on a regular basis.

We're looking for a star dev who really loves to work with a team creating the best software. Most development is in C# - at least to begin with - things have a way of changing (largely because we always want to do something new). Leadership and teamwork skills would be cool as well.

I'll keep this post up for a week or two - if you're interested - drop me a line.....

-Jamie

Free SQL Data Mining chapter from "Practical Business Intelligence with SQL Server 2005"

JamieMac — Sun, 12 Nov 2006 12:02:00 GMT

Informit.com has made available a complete data mining chapter from Practical Business Intelligence with SQL Server 2005.

The co-author of the book, John Hancock, is a Microsoft Consultant out of Canada that I've had the pleasure to meet. He visited the Analysis Services team for a bit in what's called the "SWAP" program (although it seems a bit one-sided, consultant's swap in, but we never seem to swap out, hmmmm). John had implemented a very clever application of our Association Rules algorithm to implement a link analysis solution. He took advantage of our extensive .Net programmability model inside stored procedures. It was really cool.

John and Roger's chapter is described as "This sample chapter looks at how to use some of the data mining features in SQL Server Analysis Services 2005 to perform tasks such as customer segmentation and market basket analysis. The data mining results are presented in the form of new dimensions in cubes and are used in Web applications."

I've read through it a bit and I'm impressed - it doesn't have the "too close to the product" issues that my book has, so many things have better high-level explanations, but you get the detail of how to implement the solution as well.

Help with Time Series on OLAP - another great blog with SQL Server Data Mining

JamieMac — Sat, 10 Jun 2006 21:22:00 GMT

Dejan Sarka writes about how to get "Cases from Multiple Dimensions of OLAP Cubes in the Time Series Algorithm" - an issue that has stymied many users. He also has a range of great posts on other DM and BI topics as well - worth a good read!

BI Training in LA towards the end of June

JamieMac — Mon, 24 Apr 2006 23:43:00 GMT

MSEvents is hosting a hands-on lab in June where you can learn about BI for IT professionals. From this training (according to their site) you will learn how to create