Re^2: Automating data loading into Oracle

No jokes here. This is serious. The tool is already chosen -- the client wants Shell/Perl (I am the one who is whispering "shell" and raising my volume when saying "Perl").

Fortunately, there are resources available to gather the metadata, do an assessment, study all the existing data flows, sources, transformation, and target. So, I do have some say in determining the effort. It is hard though, to try and guesstimate how much time a particular task will take, so all my numbers right now are very SWAG-ish.

--

when small people start casting long shadows, it is time to go to bed

Comment on Re^2: Automating data loading into Oracle

Replies are listed 'Best First'.
Re^3: Automating data loading into Oracle by poqui (Deacon) on Oct 25, 2005 at 20:16 UTC
When you say "resources available to gather the metadata", what exactly do you mean? Do you mean that someone is going in after you write the transformations in perl and translating that into english? Or are you simply referring to the existing process that perl is replacing? How are you capturing the metadata about the transformations you are putting the data through? In Informatica, for instance, the transformations are saved in an RDBMS, but still in their code form. If I want english translations for those transformations, I have to do that myself; plus in my Metadata Repository, I want the english explanations from the requirements document that drove the development of the code. In order for all of that to make sense across anything other than a small company, I need to capture definitions of terms, and who defined the terms and by what authority. There is also metadata generated by the process of ETL itself: When the last load ran, was it complete, was it successful, how many rows did it load, was it on time, if it wasnt complete why...	[reply]
Re^4: Automating data loading into Oracle by punkish (Priest) on Oct 25, 2005 at 22:53 UTC
When you say "resources available to gather the metadata", what exactly do you mean? Do you mean that someone is going in after you write the transformations in perl and translating that into english? Or are you simply referring to the existing process that perl is replacing? I mean the latter... understanding and "decoding" the current procedures (that someone else wrote... they are mainly DTS packages and the information inside one guy's head) into plain English. Once I recreate the ETL processes in Perl, they will all be well documented, and I will leave a conspicuous log trail. So, together that will make for good metadata anyway. -- when small people start casting long shadows, it is time to go to bed	[reply]