Any OLAP / SSAS (or Mondrian) experts?

[Kent Skinner]

Would anyone be so kind as to help me with a few questions about
measure-groups, many-to-many relationships, distinct counts, MDX etc.?

Also, any recommended books on these topics for someone new to OLAP?

Many thanks,
Kent

[kell.s...@gmail.com]

I'd love to tag along on this as well. We have a big cube at work and we have some upcoming architectural stories to take on and I've never used this before so I'd love to learn from this too!

Later,
Kell

> --
> You received this message because you are subscribed to the Google Groups "Seattle area Alt.Net" group.
> To post to this group, send email to altnet...@googlegroups.com.
> To unsubscribe from this group, send email to altnetseattl...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/altnetseattle?hl=en.
>

[Adron Hall]

If you're diving into the Business Intelligence world check out Ralph Kimball

http://www.kimballgroup.com/

As for books, some of the best material around the ideals and getting things to work out well:

http://www.kimballgroup.com/html/booksDWET.html

http://www.kimballgroup.com/html/booksDWT2.html

http://www.kimballgroup.com/html/booksDWLT2.html

If you want to dive into Microsoft's SQL Server Integration Services or SQL Server Analysis Services I personally would just say to read the books above to get the concepts, then use the "Books Online" that come with SQL Server and build out some samples, but wouldn't buy any books specifically for those two functionality.

Do NOT buy any of the Microsoft Press books, at least IMHO. Anything in them is easily found on MSDN, etc.  :)

Adron

--
Adron B Hall

Tech: http://compositecode.com
Transit:  http://transitsleuth.com
Twitter: http://www.twitter.com/adron

[Richard Turner]

As an aside, I cannot recommend http://SafariBooksOnline.com enough. It’s monthly subscription program has allowed me to save a small fortune while being able to access a vastly expanded library.

 

HTH.

 

Richard Turner

E: ri...@bitcrazed.com | T: @bitcrazed | W: http://www.bitcrazed.com

[Stevi Deter]

Ditto on the safari books online recommendation. Not cheap, but still worth every penny.

--s

[carlosk]

To learn MDX, I recommend this book: http://www.amazon.com/Fast-Track-MDX-Mark-Whitehorn/dp/1846281741

I took the BCC online classes on BI and I found them very useful in
jump starting my BI skills. From there it is a matter of reading the
blogs and practicing.
http://bellevuecollege.edu/distance/degrees/AchievBusITDev.asp

As for posting questions, I'd recommend searching on Google for: MDX
sql forum
You're more likely to get BI experts there.

Also look at the AdventureWorks samples. It includes the OLTP, ETL,
OLAP databases, and the the cube. It's a good starting place to see a
sample system.

Also try googling for: ssas filetype:pdf site:microsoft.com

I hope this helps.
-Carlos Klapp

[Erick Thompson]

I believe that I have a copy of that book. It might be an earlier edition, I will have to check. I haven't opened it in ages, so I would be happy to loan it out. 

For OLAP in general, it depends on what aspect your coming in on. Are you working on query/front end, cube design, ETL, or something else? There is a lot of material.

Erick

[Kent Skinner]

You guys are awesome - thank you! I took your advice and signed up for
Safari Online, and will check out the Kimball and MDX books.

Erick, at the moment I'm trying to determine if a data cube would make
our reports faster & easier to write. We have a cube-like structure
for most of our tables already - a row exists for every combination of
queried dimension and filter. I'm trying to figure out if using a real
OLAP system like SSAS or Mondrian and writing reports with MDX would
enable us to add new reports more easily. At the moment, each new
report involves a lot of work. On the other hand, I have no experience
with SSAS or MDX and wonder if MDX is rich enough to express some of
our more complex reports.

Anyway, I'll be reading like mad over the next couple of weeks to try
to answer this question.

Thanks again everyone,
Kent

[Adron Hall]

Are you using SSRS against the data stored?  Because if you are, I'd say you could write reports in blinding speed if you wanted to. However, you have to get an SSRS report server running and then connect to appropriate data stores - either regular SQL Server RDBMS type tables and such, or to a cube.

Also keep in mind, don't shove too much data into a cube, it isn't really Hadoop + MapReduce type capabilities there, you're just making it a little faster. Then of course, if you're stepping into serious data sizes you'll probably need to just drop SQL Server for most of it (i.e. 1+ TB and on up to the Petabyte range).

As for faster to write? Not sure, MDX is about as muddlesome and unmaintainable as SQL is, with a bad proclivity to end up designed very very top down without any sort of appropriate abstractions. In addition, there really isn't any good way to do clean TDD/BDD style dev with MDX reports against cubes. 

...there are some really weird and perverse XML frameworks that some might call unit testing frameworks, but I really can't say anything positive about those.

Another thing I'd highly suggest, vs. having a bunch of people answer without knowing your entire situation, is to just learn the basic design schemas for cubes (snowflakes, etc) and see how and where you would query data from those. You might find MDX either completely useless or super helpful. It really all boils down to what you have, how it is stored, and how you want it returned to end users.

-Adron

[Scott Koon]

Not a popular choice, but back when I dipped my toes into OLAP in 2000 I liked this MSPress book.

http://www.amazon.com/Mining-Microsoft-Technical-Reference-Professional/dp/0735612714/ref=sr_1_14?ie=UTF8&qid=1324407986&sr=8-14

"Data Mining with Microsoft SQL Server(TM) 2000 Technical Reference"

But it only covers SQL 2000 and I'm not sure how much the tech has changed since then. What I liked about it was a nice overview of the algorithms and schemas for creating data warehouses. 

Agree with Adron, MDX is a hot, sticky mess. It made my head hurt more than once.