I'm going to join with everyone else and say that if the file format is any one of the standard types (CSV, TSV, fixed width) that SSIS is the way to go. And this remains true even if you think the format might change, then you just update the SSIS packet.
I would however look outside of SSIS if the file format is genuinely nonstandard or for some reason significant preprocessing before being imported. I have for instance dealt with situations where I just needed to extract the e-mail addresses from otherwise unorganized text, or where I needed to strip out unusual characters such as an ASCII null character or bell character before importing, or where numerous lines needed to be split into header and then value to be turned into a single row in the database.
In all those cases, I turned to a more flexible programming language (normally Python, but just about any of the major ones would work).
↧