RE: data warehousing advice
- From: Xin (Sheen) Zhang [MSFT] <XinSheenZhangMSFT@xxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: Thu, 8 Jan 2009 18:04:01 -0800
You can use a concept like a Load_Id to keep track of your loads instead of
using transactions.
If some loading failed, you can easily undo the load by Load_Id.
Hope it helps
--
Shin
MS SQL Server
"Derek" wrote:
i've read the kimball books and i've created my staging tables and.
star schema (dimensions and facts). Now i need advice on my ETL (I am
using SQL Server 2005 Standard Edition).
For extract, I plan on pulling in csv files via FTP into my warehouse
server. i will then bulk insert this data into stage tables. From
the stage tables, I will transform the data and load my dimensions and
facts.
this is my question:
To keep the dimensions and facts from getting corrupted I was going to
put a transaction across the whole dimension and fact loading process
(one giant transaction). But I'm afraid of tempdb or the log file
getting huge. Is this a good idea or is there a better way? My
warehouse is set to simple recovery mode.
thanks in advance!
- Prev by Date: RE: Testing Environment + TestData + QA Setup
- Next by Date: SSAS/SSRS Report Log Tracking
- Previous by thread: RE: Testing Environment + TestData + QA Setup
- Next by thread: SSAS/SSRS Report Log Tracking
- Index(es):
Relevant Pages
|