HP3000-L Archives

February 2000, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
john edwards <[log in to unmask]>
Reply To:
john edwards <[log in to unmask]>
Date:
Thu, 17 Feb 2000 09:01:36 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (69 lines)
Hi Again
Let me add a few more words and see if I have another
winner:
1) Efficiency - Depends (As always)
It will be faster for IMAGE to add data after the HWM.
Simple reason is that it will not need to maintain the
deleted entry chain. There will also be less physical
IO. If there are 10 entries per block and you have
deleted every 10th record, then adding 100,000 entries
will require the processing of 100,000 blocks. Adding
100,000 entries after the HWM would require the
processing of only 10,000 blocks. The importance of
blocking is not so great now days.
You can bring about this state by:
a) reorganising your detail set after all those
deletions so that the HWM (and the last entry) is
positioned at the number of entries in the set, ie no
deleted entry chain.
b) set the HWMPUT option to force IMAGE to ignore the
deleted entry chain and always place new entries after
the last entry in the set.
c) delete all the entries, sort them and put them
back.
2) HWMPUT
Deleting records will still create a deleted entry
chain and you will end up with a lot of empty space at
the beginning of the dataset. Eventually you will run
out of space (even with DDX) and be forced to do a
dataset reorganisation. A serial read of the dataset
will also be reading a lot of empty blocks at the
start of the set, which will slow down that type of
processing.
3) Depends
If you can aford the down time, then reorganise your
dataset after any large batch of deletions (or
additions).
Regards
John Edwards



--- Doug Werth <[log in to unmask]> wrote:
> >
> >
> > Doug,
> >
> > But Steve Bruno ended up with a _lower_ HWM not a
> higher one in the last
> stage.
> > How does that happen?
> >
>
> Whoops!  Re-read the original post. You are
> absolutely right, it did go
> down.  I thought it had gone up. Sorry for the
> mis-information.
>
> Doug.
>
> Doug "cleaning his contacts" Werth     Beechglen
> Development Inc.
> [log in to unmask]
> Cincinnati, Ohio
>
__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

ATOM RSS1 RSS2