Data warehousing appliances...

[k1ng87]

Does anyone have any experience with these, such as Netezza, Datallegro, Teradata, Greenplum, etc... We're starting a data warehousing project to consolidate our data, normalize it, and setup a standard data model.

Right now, we draw data from multiple source systems and stick the feeds(transformed by Informatica) into a data mart(we have an extensive amount of these) and have a different data model for each data mart.

The problem with this is that that same source system could be feeding multiple data marts but being transformed differently. This makes it difficult for us to figure out which is the "Truth" I guess i would call it. Then multiply that by the 200-300 terabytes of data we have (yes terabytes), it makes it very difficult to properly use our analytical tools to their fullest potential.

I've uploaded a mock up of how the data flows.

http://www.sendspace.com/file/95t6ad

 

[Hans63]

Sorry, but I have no knowledge of any of those.
From your description, I should say you definitely need to conform the data.

Beware that you need the business to define what the 'truth' is for your situation. Otherwise you will remain stuck in this swamp of different truths. When I was working on a datawarehouse for a large bank/insurance company I encountered the same situation. We wanted to get to 1 definition of the truth, one member of the board of directors told us: No, as long as the organisation does not, by itself come with an initiative to define 1 version of the truth, you must be able to provide many version of the truth. That is a technical problem, so it is solveable. If we do otherwise, we are pushing our technology onto the business. Which is the same mistake ICT has made for several years.