Data Warehousing Community Forum
September 20, 2019, 06:56:15 pm

Pages: [1] 2 3 ... 10
 1 
 on: August 06, 2019, 10:16:45 pm 
Started by topcat - Last post by topcat
I have a series of dimensions and all are related hierarchically, related in the source system that is.  what is the best way to model this?  for example, i have these dimensions

sales agent dim
store dim
client dim
master client dim

these dimensions are all related: sales agent belongs to store, store to client, client to master client

some facts connect at the agent level, some at the store level. 

my reporting tool uses hierarchies to allow user to aggregate at various levels.  so one could roll up sales at the store level, client level, or master client level.   the tool likes the hierarchy data to come from a single table.   the table would include agent, store, client, and master client.

so my first thought is to just put all the data in the agent dimension.  connect agent dim to my facts and expose all the levels of the hierarchy from the agent dim.   

but there are also some facts that connect at the store level, so i would also have a store dim that includes the hierarchy above it.  that would be store, client, master client.  so store dim has one level less than agent dim.

now am thinking that there would be a bunch of replicated data as agent will contain store and above, store dim will contain client and above. 

all of this i am ok with, but some reporting tools only like to source an attribute from one place.  for example, the client name will exist in the agent dim as well as the store dim.  so when someone drags the client name to the reporting pallet, the tool has to know where to get it from.  for us sql writers, its pretty easy.  not so easy for a reporting tool.

so my question is whether to keep the whole hierarchy in the lowest level dimension, or simply join dimensions together.  ie have fk of one dim in the other.  i dont like this idea, but is certainly cleaner.

also, all of these dimensions are type 2.  so one added complexity is if we store the whole hierarchy in the agent dimension, each time the higher dimensions get modified, then every child dimension record will have to change.   ie if we change the master client description, then every agent record that contains that master client will get updated.  new record created for every agent.

thoughts?

 2 
 on: March 29, 2019, 07:57:12 pm 
Started by welschii - Last post by welschii
Hi,

I'm trying to put together a business case (or product vision) to convince a department's director to give the go ahead to move their data into our corporate infrastructure and ultimately EDW. The problem is the department are currently executing a project for a new line of business application, or system of record, for their operation. Unfortunately, the supplier involved has recommended to them that they design/build a data warehouse for them in addition to the line of business application. To be honest the team in the department involved in executing the project is too close the supplier. So, there are competing forces at play. Whilst they are free to go down this route, however ill-advised, I need to put together a water-tight business case that explains the benefits in layman's terms of why we should have the data moving into the corporate infrastructure also to avoid a silo situation.

Some bullet points I've put together so far:

* Remove barriers to corporate BI/Analytics tools
* Remove barriers to access data both within the department, and across departments in the business
* Pathway to single source of truth from system of record - can trust the data

Does anyone have any advice or suggestions in regard to the business case?

Thanks,

MJ



 3 
 on: March 29, 2019, 02:08:06 pm 
Started by Akriti - Last post by divyapriya
Can anyone suggest me ideas on my ielts coaching classes.

 4 
 on: March 25, 2019, 11:58:27 am 
Started by soumitasai - Last post by soumitasai
Credo Systemz offers an inclusive Big Data training in chennai. The extensive practical training provided by Big Data training institute in chennai equips live projects and simulations. Such detailed Big Data course has helped our students secure job in various MNCs. The trainers at Credo Systemz are subject specialist corporate professionals providing in-depth study in Big Data course in Chennai. Participants completing the Big Data certification have plethora of job opportunities in the industry.

For Futher details, Call us :+91 9884412301 / 9600112302

Pages: [1] 2 3 ... 10