SSAS - Lines, Measures and Cubes Ohh My!

  • I know very little about SSAS, so little you might as well assume I know nothing but some buzzwords. I'm looking at trying to use SSAS to solve performance issues our company has with financial reports. We currently have a SQL Server(2005) administered traditional relational database that's just under 300GB with hundreds of millions of rows, soon to be measured in the billions a few years from now based on current growth projections,

    There are times when the financial reports our accountants and big wigs want to see are hit on the system over all because of the amount fo data that must be aggregated. I'm thinking this problem has Data warehouse written all over it which leads me to my first question.

    1) What exactly is an Analysis Services thing? In the traditional RDBMS side we have a database with 0 to many tables, views, stored procedures and so on. But with SSAS its not as clear from what I've been able to read so far. For example if i want to take my traditional Relational DB data and use Analysis services do I use my current DB as the source or does SSAS require something different and so I would export data from my relational DB but I would also then convert that into something that Cubes in SSAS would use?

    2) Is a Data Mart the same thing as a Data Warehouse as far as SQL Servers concerned?I use to think I knew what a Data Mart was but then I saw software vendors a few years back start using that term to describe their own traditional RDBMS based tables that were nothing but non-normalized data, aggregated to commonly requested value. There were no measures or a Cube just pre-aggregated values .

    As far as SSAS goes is a Data-mart and a Data Warehouse the same thing?

    Does an Analysis Services thing require traditional RDMS to serve as a data store or source?

    If one wants to improve reporting on large aggregated values and they know that the results on said data would note real time but on as of whenever the last time was that the cue was refreshed, would SSAS be a solution and would this require doing something to the DB like getting a copy and then converting it or does SSAS just pull what it needs from the relational database?

    I know I can get more details and probably answers to many of these by reading the info on each but I 'd like to determine as soon as possible if SSAS is a possible option or just a waste of time.

    Thanks

    Kindest Regards,

    Just say No to Facebook!
  • YSLGuru (2/26/2011)


    I know very little about SSAS, so little you might as well assume I know nothing but some buzzwords. I'm looking at trying to use SSAS to solve performance issues our company has with financial reports. We currently have a SQL Server(2005) administered traditional relational database that's just under 300GB with hundreds of millions of rows, soon to be measured in the billions a few years from now based on current growth projections,

    There are times when the financial reports our accountants and big wigs want to see are hit on the system over all because of the amount fo data that must be aggregated. I'm thinking this problem has Data warehouse written all over it which leads me to my first question.

    1) What exactly is an Analysis Services thing? In the traditional RDBMS side we have a database with 0 to many tables, views, stored procedures and so on. But with SSAS its not as clear from what I've been able to read so far. For example if i want to take my traditional Relational DB data and use Analysis services do I use my current DB as the source or does SSAS require something different and so I would export data from my relational DB but I would also then convert that into something that Cubes in SSAS would use?

    2) Is a Data Mart the same thing as a Data Warehouse as far as SQL Servers concerned?I use to think I knew what a Data Mart was but then I saw software vendors a few years back start using that term to describe their own traditional RDBMS based tables that were nothing but non-normalized data, aggregated to commonly requested value. There were no measures or a Cube just pre-aggregated values .

    As far as SSAS goes is a Data-mart and a Data Warehouse the same thing?

    Does an Analysis Services thing require traditional RDMS to serve as a data store or source?

    If one wants to improve reporting on large aggregated values and they know that the results on said data would note real time but on as of whenever the last time was that the cue was refreshed, would SSAS be a solution and would this require doing something to the DB like getting a copy and then converting it or does SSAS just pull what it needs from the relational database?

    I know I can get more details and probably answers to many of these by reading the info on each but I 'd like to determine as soon as possible if SSAS is a possible option or just a waste of time.

    Thanks

    Others will clarify further, but this is my understanding

    1) Analysis services is the engine for producing answers to business intelligence that might not easily be answered in conventional sql queries (or take a long time to produce). An Analysis services deployment consists of using cubes and these cubes are typically based on standard RDBMS tables, but they are highly denormalized for improved read speeds. The cubes contain measures (aggregatable data) and dimensions (the what/where/who type filters).

    2) I interchange the terms "datamart" and "datawarehouse", but datamart infers some smarts you can get out of the system , where as to me a data warehouse is just a lot of data, yet to be used and converted into information.

    Depending on how large your cubes end up being, processing can be done on a pretty regular basis (even just before the financial reports are run if you like), meaning the data is as up to date as you want it to be.

    You dont need to make a copy of the database as SSAS doesn't typically write to the database, but from performance it can be better to have heavily reported on information separate from where the write transactions are happening.

    HTH/

  • 2) I interchange the terms "datamart" and "datawarehouse", but datamart infers some smarts you can get out of the system , where as to me a data warehouse is just a lot of data, yet to be used and converted into information.

    Incorrect!! The data mart and Data warehouse are two all together different entities.

    Raunak J

  • 2) I interchange the terms "datamart" and "datawarehouse", but datamart infers some smarts you can get out of the system , where as to me a data warehouse is just a lot of data, yet to be used and converted into information.

    Incorrect!! The data mart and Data warehouse are two all together different entities.

    I wish you had explained why you think they are all together different entities. I don't agree, but without knowing why you think that a response is difficult.

    The terms are not interchangeable, however. The data warehouse is the sum total of the data marts that may exists. There's also a new suggested term for data mart, something like business object process. To a beginner, however, I would simply say that a data warehouse is composed of one or more data marts. If more than one, the data has been conformed so that the terms for each data mart are consistent with the same terms when used in another data mart.

  • RonKyle (2/28/2011)


    2) I interchange the terms "datamart" and "datawarehouse", but datamart infers some smarts you can get out of the system , where as to me a data warehouse is just a lot of data, yet to be used and converted into information.

    Incorrect!! The data mart and Data warehouse are two all together different entities.

    I wish you had explained why you think they are all together different entities. I don't agree, but without knowing why you think that a response is difficult.

    The terms are not interchangeable, however. The data warehouse is the sum total of the data marts that may exists. There's also a new suggested term for data mart, something like business object process. To a beginner, however, I would simply say that a data warehouse is composed of one or more data marts. If more than one, the data has been conformed so that the terms for each data mart are consistent with the same terms when used in another data mart.

    cool

    you learn something new everyday.

    🙂

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply