Hi, I have a Data Integration Studio project that we use to do ETL for a survey that is fielded on a regular basis. Each round, we will have a nearly identical DIS job, but not *completely* identical, extracting data from one place and putting it in another. (It will be identical in how it works, but there may be updates to the logic or the objects each round, and the physical location of the source and destination libraries varies per round.)
The way we plan to implement this is to have a Metadata folder for each round, with the jobs and the source and target libraries defined in it. To deploy a new round, we will copy one round to a new folder named for that round, rename (in Management Console) the libraries (which will be automatically renamed to something like "Library Name (1)") to something that indicates what round they are for, and then move forward with any updates to the programs or metadata. DIS/Management Console seems to handle the metadata objects well there.
The question here is twofold: one, is this a good idea (or the best practice); and two, will there be any difficulty with, for example, our libraries having the same SAS libname but pointing to different folders?
For a sense of scope, we have about 80 jobs in this project, each of which has quite a few steps, sometimes including user written code (we avoided it when we could, but it's not always avoidable).
We would love to find a solution that didn't require multiple copies of code existing, but don't see a way to do it that way - we need to have multiple rounds active at once (both because we could have two rounds *in field* at once, but also we typically continue to refine the ETL after the round is finished fielding as we find new errors in the survey implementation or the ETL).
Thanks!
... View more