Stuart

David has a point.  What we do is a writeseq to a disk file and once
this is completed the UVBasic program executes a unix routine to do an
ftp of the file to another position.  So if the ftp fails, the file
still exists.  The BASIC  Program also deletes the previous file as a
process of self maintenance

Regards

David Jordan

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Hona, David S
Sent: Friday, 6 February 2004 2:32 PM
To: 'U2 Users Discussion List'
Subject: RE: [UV/AIX] gzip data to file?


Stuart,

Ah, one disadvantage, of doing the 'all-in-one' approach is that if one
step fails, you have to start again. A real problem if the output takes
hours to produce and it is part of database update phase too. :-(

For example, if your compression step failed, due to disk space running
out
- will you have to run everything again?

Our sequential file generation can be re-run also, as it doesn't do any
updating itself. This is handy also, in case there is a data issue.

Nothing worse that having to restore data, just to you start again. 

We're experienced about every problem you can think of, hence we're
broken up in to 'repeatable' steps. Works for us.

Regards,
David


-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
Behalf Of Stuart Boydell
Sent: Friday, February 06, 2004 1:06 PM
To: U2 Users Discussion List
Subject: RE: [UV/AIX] gzip data to file?


> If you can, allow the spooler subsystem to do the work.
Cheers Ray, that seems like an efficient way to do it.

> We create a 80MB sequential file (plus a few small ones) with a UV 
> BASIC
program.
Thanks David, I'm thinking something similar but that outputing to a
device or spooler with a processing script might be the way to go for
us. I'm trying to keep the IO to a minimum if possible. At least if the
spooler is doing any it's at a lower (faster?) level than writeseq.

> why the aversion to writing the data to disk?
I don't think there's a need to write it until the data is in its final
shape. The system already has enough IO to deal with and in theory
should be able to keep the data in memory, which may or may not use disk
caching, but which should be a lighter touch on the system. S

-- 
u2-users mailing list
[EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users

-- 
u2-users mailing list
[EMAIL PROTECTED]
http://www.oliver.com/mailman/listinfo/u2-users

Reply via email to