Subject: | |
From: | |
Reply To: | |
Date: | Fri, 6 Mar 1998 12:54:21 -0600 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
At 03:43 PM 3/5/98 -0500, Steve Donnelly wrote:
> The circular (CIR) files
>are being used as archival logs. As files (such as EDI) are received/
>transmitted/etc, they are appended to the circular log files. Whenever we
>need to research something, it's a simple matter of scanning the circular
>file for the data in question.
>
>The beauty of circular files is they never 'fill up'. Our applications can
>write to the files without danger of writing beyond the file limit.
It occurs to me that when the volume of logged data is substantial, that
there exists another mechanism, though seemingly rarely used, which might
be appropriate. The user logging system, which most people only think
about in terms of Image logging, is actually designed as a general purpose
logging tool. The WRITELOG() intrinsic allows an application to write to a
log identifier, as I understand it even the logID used by Image logging.
And since user logs can now roll automatically from one file to another
just like system logs do, they don't fill up either. (Log files don't
overflow, they just roll over. ;)
Caveat emptor: I've thought of this several times before, but I've never
actually used this technique, so I don't really know what issues might
arise if it's implemented.
--
Jeff Woods
[log in to unmask] at Tivoli Systems
[log in to unmask] at home [PGP key available here via finger]
Haiku by Rahul Sonnad:
There is a chasm
of carbon and silicon
the software can't bridge.
|
|
|