On Thu, 20 Feb 2003, Wiggins d'Anconia wrote:

> Date: Thu, 20 Feb 2003 19:08:51 -0500
> From: Wiggins d'Anconia <[EMAIL PROTECTED]>
> To: John Baker <[EMAIL PROTECTED]>
> Cc: Colin Johnstone <[EMAIL PROTECTED]>, [EMAIL PROTECTED]
> Subject: Re: Text file or database
>
> John Baker wrote:
> > A work-mate of mine and  were having the same discussion recently.
> > I'm not a big fan of dbases, but maybe that's because I don't have
> > to be...yet.
> >
> > A couple things that could factor when considering dbases:
> > 1. speed
> > 2. additional middleware technology
> >
> > Is the speed of accessing dbase info more efficient than
> > text files with regexes?
> >
> Much for large amounts of data.
>
That was the implicit conclusion my work-mate and I came to, basically.

> 200 e-mail addresses, probably not.
>
Agreed.

> > Also when you implement dbase (middleware) code, you've
> > added an additional layer of complexity.  And another
> > technology to learn, and I'm all about adapting to
> > technology, but I'm more a fan of the KISS principle. =)
> >
>
> Ah the beauty of DBI. Write the Perl code once in DBI then just change
> one line of code no matter what OS or DB you are using.
>
> To the original question....
>
> 1 word. Network.  What happens when all of a sudden your data has to be
> stored on a different server, but your application stays put.  Sure
> there is NFS, but then your security could be shot to hell and your
> speed really drops. Or multiple users are hitting the resource at the
> same time?  Have you built in all of the proper locking, etc. Or
> multiple local users have to access the same resource with the proper
> locking, etc. better make sure those permissions are correct, etc.
>
1 word. CVS. One app aggregates the data feeds it into CVS and
other machines pull down the tree.


> Point being there is a reason why you chose Perl in the first place,
> because it was probably written by someone who knew more than you, or at
> the very least more people had the chance to look at it and work on it
> so it is far more likely to catch all of the little things that take the
> most time to handle and that end up coming back to bite you later.  If
> flat text files is the way to go for you,

> say because your boss who
> can't seem to work his e-mail wants (thinks he wants) to make a change,
> then by all means check out DBD::CSV, etc.
>
Hmm, no comment, my boss might be subscribed to this list. =)


> Then later when he leaves
> for the weekend, install a real DB, change a couple lines of code, and
> when he comes back, make up a story about insecurity of the old system
> and why you should get a raise ;-)...
>
Point taken. There are some situations where DBs, particularly DBI,
are useful; others, they're not.


> Always use a DB? No, those 6 configuration values you don't want to hard
> code using constants that only change every 6 months, sure, throw them
> in a text file and forget about it. Production data...well hopefully not
> mine.
>
Or if you have 500+ production devices and want to store each of
their configs in seperate text files instead of adding an additional
layer of complexity with a DB  (http://www.shrubbery.net/rancid/).

jab


-- 
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to