Comments on: Making rapid prototypes for data warehouse ETL jobs http://www.datamartist.com/making-rapid-prototypes-for-data-warehouse-etl-jobs Reduce cost with self serve data transformation Thu, 02 Feb 2012 18:04:29 +0000 hourly 1 http://wordpress.org/?v=3.1.3 By: James Standen http://www.datamartist.com/making-rapid-prototypes-for-data-warehouse-etl-jobs/comment-page-1#comment-1518 James Standen Thu, 24 Sep 2009 19:11:18 +0000 http://www.datamartist.com/?p=3022#comment-1518 Nope- that percentage is from my experience, and experiences reported to me by others. I worked for about 6 years on various projects, and kept budget figures for all of them (always trying to improve the estimation side)- overall the ETL was consistently in that range. I've heard others offer the same figure, or concur, but no hard data. Which is a pity, because I was always facing exactly what you described- management was always willing to pay for the flash on the front end, but wanted to know why the data transformation was so expensive. The problem, of course, with trying to compare from project to project is that they are all different- very complex front ends in situations where there is only one source system, data quality is good and a robust master data process has been in place will change the ratio significantly, for example. Nope- that percentage is from my experience, and experiences reported to me by others. I worked for about 6 years on various projects, and kept budget figures for all of them (always trying to improve the estimation side)- overall the ETL was consistently in that range. I've heard others offer the same figure, or concur, but no hard data. Which is a pity, because I was always facing exactly what you described- management was always willing to pay for the flash on the front end, but wanted to know why the data transformation was so expensive.

The problem, of course, with trying to compare from project to project is that they are all different- very complex front ends in situations where there is only one source system, data quality is good and a robust master data process has been in place will change the ratio significantly, for example.

]]> By: Robert Folkerts http://www.datamartist.com/making-rapid-prototypes-for-data-warehouse-etl-jobs/comment-page-1#comment-1515 Robert Folkerts Thu, 24 Sep 2009 13:09:07 +0000 http://www.datamartist.com/?p=3022#comment-1515 Do you have numbers or references to back up your 70% figure? My management is focused on the features of the analytics server and is giving minimal resources to ETL. I would love to show them industry averages. Do you have numbers or references to back up your 70% figure? My management is focused on the features of the analytics server and is giving minimal resources to ETL. I would love to show them industry averages.

]]>