In this installment I'm going to cover where we are at, the good and the bad, and set it up for where we're going.
I had hoped to jump right in to the next installment to focus on the technical aspect of where we have been with the EDW but got sidetracked by production problems.
Our Banner data warehouse is populated using Informatica as the ETL tool. I've been impressed with the performance the tool has for moving data around. I don't know how much the initial design of the ETL process came from the consultants that were used during construction but we ended up with the following architecture:
BANNER ---(Extract)---->Staging DB-----(Transform)---->Staging DB---(Load)---->EDW
When we began our project to implement Banner back in 2000 or so we also began a parallel project to build an EDW. Given the size and scope of the Banner implementation it was decided that the EDW project should be outside the ERP project administration. Thus we ended up with 2 new departments, one to implement Banner and one to implement an EDW right along side of the ERP implementation. Once the Banner and EDW projects were complete the Banner project department was disbanded, however the EDW department continued on, basically only the funding source changed.
One of our greatest challenges in the past few years here at the University of Illinois has had to do with the nightly population of our locally grown data warehouse. It currently takes 14 hours or so for the ETL to run each day. It starts at midnight so it usually finishes around 2 PM. Right, not a real people pleaser with that kind of delivery time. I manage the DBA team here for our enterprise applications which includes both our Banner system and our Enterprise Data Warehouse. Complicating the whole technical set of issues has been some political/organizational issues.
Recent comments
4 days 21 hours ago
3 weeks 4 days ago
6 weeks 3 hours ago
8 weeks 1 day ago
9 weeks 5 days ago
9 weeks 6 days ago
10 weeks 7 hours ago
11 weeks 2 days ago
11 weeks 6 days ago
13 weeks 4 days ago