Consolidating data using datamarts

posted by | Leave a comment

Also, Do Teradata views support SOA, and other newer technologies, if so to what extent, i.e.

can Teradata views be a full partner in the future enterprise architecture as much as Data Marts. Baz77 Yes TD can replace the Data Marts etc, and probably do it well in terms of performance.

TD Technology is unique, and well worth considering, however so are some of the things associated with the company unique !!! Regards, Ashwin" id="ctl00_m_m_i_ctl00_gr_ctl03_bestanswerbody" class="textarea-bestanswerhidden" name="bestanswerbody" answerbody Id="1003704" / We have made very good business over the course of the last couple of years helping organizations consolidate their data marts onto a single, centralized platform=2E These organizations normally have three main motivations for consolidating their data marts - (1) Redundant data =3D redundant costs (redundant hardware and licenses, increased maintenance costs, etc=2E) (2) Redundant data =3D an inconsistent data asset, with serious consequences for data quality - particularly important in the light of world-wide regulatory and compliance initiatives=2E (3) Cross-functional analysis (e=2Eg=2E comparing sales with inventory) is typically difficult or impossible for organizations with a very distributed analytical architecture=2E There are other considerations - for example, there is a lot of academic research which demonstrates that it is easier to secure a single data repository than it is to consistency enforce a single security policy across multiple data repositories=2E Many organizations have built analytical applications that they want to continue to use and that are designed around data mart schemas; we typically recommend that these organizations migrate the data in these data marts to an application-neutral EDW schema and build database views on top of this new schema to simulate the old data mart structures so that these applications can continue to function normally with minimal modification=2E Some organizations elect to just "lift-and-shift" the data mart schemas onto a centralized platform (we call this "forklift migration"), normally in the interests of securing a very rapid migration; these organizations secure TCO benefits but may still have issues with data quality and consistency=2E Teradata is designed and optimized exclusively for BI workloads; our view implementation is extremely efficient and we have a number of features that can be used to further enhance performance of specific workloads, e=2Eg=2E join and aggregate join indexes (like materialized views only better), partitioned primary indexes, etc=2E, etc=2E I don't propose to discuss these further here for fear of turning this into a naked sales pitch=2E We also support SOA: we have a J2EE-compliant application framework (Teradata Application Platform or TAP) designed to facilitate SOA; we are re-building our own analytical applications atop this framework so that in future we can market analytical services as opposed to entire applications and we have recently entered into a partnership with SAP to support their Enterprise Services Architecture (ESA) SOA-play=2E If you would like more information or Teradata collateral (white papers about data mart consolidation, etc=2E) please contact me on 44 7788 190 718 and I will put you in touch with a local Teradata representative=2E Alternatively, if you visit www=2Eteradata=2Ecom and click the "Resources" link you will find lots of white papers and other material=2E Thanks and good luck with your initiative whatever path you choose to follow, Martin Willcox=2E" id="ctl00_m_m_i_ctl00_gr_ctl07_bestanswerbody" class="textarea-bestanswerhidden" name="bestanswerbody" answerbody Id="1003813" / Firstly let me remind you all that I work for Teradata and so may be biased=2E=2E=2E :-) We have made very good business over the course of the last couple of years helping organizations consolidate their data marts onto a single, centralized platform=2E These organizations normally have three main motivations for consolidating their data marts - (1) Redundant data =3D redundant costs (redundant hardware and licenses, increased maintenance costs, etc=2E) (2) Redundant data =3D an inconsistent data asset, with serious consequences for data quality - particularly important in the light of world-wide regulatory and compliance initiatives=2E (3) Cross-functional analysis (e=2Eg=2E comparing sales with inventory) is typically difficult or impossible for organizations with a very distributed analytical architecture=2E There are other considerations - for example, there is a lot of academic research which demonstrates that it is easier to secure a single data repository than it is to consistency enforce a single security policy across multiple data repositories=2E Many organizations have built analytical applications that they want to continue to use and that are designed around data mart schemas; we typically recommend that these organizations migrate the data in these data marts to an application-neutral EDW schema and build database views on top of this new schema to simulate the old data mart structures so that these applications can continue to function normally with minimal modification=2E Some organizations elect to just "lift-and-shift" the data mart schemas onto a centralized platform (we call this "forklift migration"), normally in the interests of securing a very rapid migration; these organizations secure TCO benefits but may still have issues with data quality and consistency=2E Teradata is designed and optimized exclusively for BI workloads; our view implementation is extremely efficient and we have a number of features that can be used to further enhance performance of specific workloads, e=2Eg=2E join and aggregate join indexes (like materialized views only better), partitioned primary indexes, etc=2E, etc=2E I don't propose to discuss these further here for fear of turning this into a naked sales pitch=2E We also support SOA: we have a J2EE-compliant application framework (Teradata Application Platform or TAP) designed to facilitate SOA; we are re-building our own analytical applications atop this framework so that in future we can market analytical services as opposed to entire applications and we have recently entered into a partnership with SAP to support their Enterprise Services Architecture (ESA) SOA-play=2E If you would like more information or Teradata collateral (white papers about data mart consolidation, etc=2E) please contact me on 44 7788 190 718 and I will put you in touch with a local Teradata representative=2E Alternatively, if you visit www=2Eteradata=2Ecom and click the "Resources" link you will find lots of white papers and other material=2E Thanks and good luck with your initiative whatever path you choose to follow, Martin Willcox=2E alas, there is no truth to the idea that simply placing a set of views over a 3NF model to present it as a dimensional model buys you the same benefits as implementing a broad set of features of dimensional models=2E A couple of TD folks here have suggested that 'everything can be done in views' or 'almost everything' and the facts are that if the underlying model is 3NF then no matter how you want to present it, it does not support many important features achieved by things such as generating internal keys and muliple levels of summary inherently supported in the design=2E I've published plenty of free material on www=2Epeternolan=2Ecom about the benefits of dimensional models=2E TD folks are now saying that the database optimiser understand dimensional models and if that means building the underlying data as dimensional that's fine=2E But merely building a set of views over the top of a 3NF model and calling it dimensional merely indicates a profound lack of understanding of the other benefits not achieved by this approach=2E Best Regards Peter Nolan Original Message: -----------------" id="ctl00_m_m_i_ctl00_gr_ctl09_bestanswerbody" class="textarea-bestanswerhidden" name="bestanswerbody" answerbody Id="1003831" / Hi Baz77, alas, there is no truth to the idea that simply placing a set of views over a 3NF model to present it as a dimensional model buys you the same benefits as implementing a broad set of features of dimensional models=2E A couple of TD folks here have suggested that 'everything can be done in views' or 'almost everything' and the facts are that if the underlying model is 3NF then no matter how you want to present it, it does not support many important features achieved by things such as generating internal keys and muliple levels of summary inherently supported in the design=2E I've published plenty of free material on www=2Epeternolan=2Ecom about the benefits of dimensional models=2E TD folks are now saying that the database optimiser understand dimensional models and if that means building the underlying data as dimensional that's fine=2E But merely building a set of views over the top of a 3NF model and calling it dimensional merely indicates a profound lack of understanding of the other benefits not achieved by this approach=2E Best Regards Peter Nolan Original Message: ----------------- Given the vertical you are in then TD would be a good solution for you.

consolidating data using datamarts-24consolidating data using datamarts-24consolidating data using datamarts-31

MW" id="ctl00_m_m_i_ctl00_gr_ctl23_bestanswerbody" class="textarea-bestanswerhidden" name="bestanswerbody" answerbody Id="1004480" / OK=2E Scenario (1) seems to be business-as-usual for you so I won't comment on it any further=2E Scenario (2) is associated with some opportunities and also some challenges=2E I'll try and deal with these generically and independent of your choice of platform technology before I pitch Teradata (!

Further, would it be desireable to have both TD views and data marts depending on the situations/projects etc.

Many thanks" id="ctl00_m_m_i_ctl00_gr_ctl21_bestanswerbody" class="textarea-bestanswerhidden" name="bestanswerbody" answerbody Id="1003952" / Hi All, Thanks for your builds, I should make clear that its not the aim to consolidate existing data marts into TD.

Hi All, I am currently evaluating an Enterprise Architecture design which includes among other components, a Teradata DW, an Oracle ODS, and Oracle Data Marts. It has been claimed that Teradata views can replace the Data Marts.

I would like to find out the views performance versus the Data Marts.

Leave a Reply

Meet and sex chat manila chatroom