Usually, ODS has more details of DW, but covering much less time range than
Also, some reporting might need to query data from ODS instead of looking at
aggregated data from DW.
For example, a retail store need detailed information of supply chain
information from different sources for half year to track back any problems
found in the chain.
And in their DW, they only care about the stock and pricing information in
past 5 years. Then, keeping 5 years of detailed transactions integrated from
different data source is not designed to fit into DW. They need an ODS in the
middle to support their need.
Bottom line, ODS is optional to build an DW.
Hope it helps.
MS SQL Server
> I really need your input for the following architecture data
> I was suggested to create an ODS after staging the data and then
> create a dimentional DB
> (sources >> staging >> ODS >> DW (DataWarehouse))
> However, I don't see the usefulness of ODS.
> My argument
> 1. whatever you want to get in ODS, I can get it from DW.
> 2. it takes more time and resource to load An additional system.
> Here are my preference in the order
> 1. sources > staging >> DW
> 2. sources > (staging/ods combine) >DW
> 3. source > staging >DW and optional (staging >> ODS)
> (incase you want to do a daily run for dw and weekly run for the
> is there anyone seeing the same way as I do?
> I strongly believe in the old days (10 yrs ago) ODS was created
> because of the software and hardware issues.
> Nowaday, there are many ETL apps and hardware are cheap. we don't need
> extra step in ODS.
> my background:
> sucessfulLy developed multiple DW's with no ODS
> extremely good with SQL AND SSIS.