About Databases...

[Simon John]

Tom Willis wrote:

[snip]
> Whoa, you are asking alot. Without knowing anything about your
> requirements except what was mentioned in your post. I would say you
> would quite possibly want the functionality of a relational database.

I'm not sure I agree with that.

If the data is likely to be large, binary and floats/integers, I'd
probably say a database is sub-par. I'd probably go with a filesystem
approach - like a directory structure (not just a single file) that
your can do an os.walk() on.

You could have some unique directory naming convention (maybe based on
oil facility name, plus a timestamp) and within each of these
directories store your binary as one file, and your statistics
(integers/floats) as another, maybe in some standard-ish format, such
as CSV, .ini (ConfigParser style) or heaven's forbid, even XML (using
sgmlop).

You'd have to research an optimal filesystem based on your needs, for
example ext3 is good for writing large files and Reiser is good for
reading small files (I think that's how it goes).

But otherwise, yeah you can connect to MySQL (using MySQLdb/adodb),
PostgreSQL (using psycopg), MSSQL, Oracle etc. using Python.