RESUSEDLT(*YES) - Performance Overhead !!??
mahabala hoskere
performance on write/delete of records? What we have is a large no. of
files having around 1 million records with the write/delete activities on
these files being very frequent. Hence deleted records occupy a lot of
space. One solution was RGZPFM regularly. If we use RGZPFM, it will take
hours of downtime to build accesspaths for about 20-25 logical files for
each PF. The other obvious solution is REUSEDLT(*YES) on PF. But we do not
know whether it will affect peformance searching for the first deleted
record for a write statement. AS/400 documentation doesn't mention anything
on this. Any help will be appreciated.
-**** Posted from RemarQ, http://www.remarq.com/?b ****-
Real Discussions for Real People
Tim
never noticed a performance problem. The documentation does state that the
system will attempt to use a deleted record on a write operation but may
not. So I believe the system has been coded to not spend too much time
trying to find deleted records.
You have to weigh the small, on-going cost of REUSEDLT on each write with
the significant cost of periodic downtime with RGZPFM.
mahabala hoskere <hsmah...@hotmail.com> wrote in message
news:BV4E3.266$664.109969@WReNphoon3...
> Hello All..I want to know whether REUSEDLT(*YES) on the PF will affect
> performance on write/delete of records?What we have is a large no. of
> files having around 1 million records with the write/delete activities on
> these files being very frequent. Hence deleted records occupy a lot of
> space. One solution was RGZPFM regularly.If we use RGZPFM, it will take
> hours of downtime to build accesspaths for about 20-25 logical files for
> each PF.The other obvious solution is REUSEDLT(*YES) on PF. But we do not
> know whether it will affect peformance searching for the first deleted
> record for a write statement.AS/400 documentation doesn't mention anything
Juper Wort
worse if there are very few deleted records, as you suggest. Also watch out
if you have any programs that read the PF sequentially, as the order if
records will be lost, ie newest not at the bottom.
mahabala hoskere wrote in message ...
Jack
based on it) with REUSEDLT(*YES). The file is used to be processed batch but
one interactive apps access the file through DDM for occasional querying.
Since we started to remove old info we set on the delete recs recycling (no
chance between RGZPFM and REUSEDLT for us) but we don't have bad effects on
performance (expecially in write/delete record as you ask).
Surely it need to be properly reorganized due to indexes degradation (if any
exists). But it depends on the purpose it is for (sequential vs random
access).
Consider to implement EVI (SQL only) for performance improvements.
Bye.
mahabala hoskere <hsmah...@hotmail.com> wrote in message
BV4E3.266$664.109969@WReNphoon3...
hs mahabala
What is EVI? How does it help improve performance?
Thanks,
Harry
Jack <aaa@zzz> wrote in article <37e0...@news2.prserv.net>...
Joel S. Mueller
EVI in AS/400, you should find several hits. For example -->
http://www.as400.ibm.com/developer/bi/evi.html
Joel S. Mueller
hs mahabala <hsmah...@hotmail.com> wrote in message
news:01bf0345$ed98bce0$480394ad@cop3_95w.ban.india.bosch.com...
Jack
via SQL CREATE ENCODE VECTOR INDEX..... from V4R3.
Please look at
http://www.as400.ibm.com/developer/bi/evi.html
or
http://publib.boulder.ibm.com/cgi-bin/bookmgr/BOOKS/QB3AQ902/5.14
Bye.
hs mahabala <hsmah...@hotmail.com> wrote in message
01bf0345$ed98bce0$480394ad@cop3_95w.ban.india.bosch.com...
> Hello,
> What is EVI? How does it help improve performance?
> Thanks,
> Harry
Kent Milligan
take the DB2 UDB for AS/400 SQL & Query Performance class. You can find a link
to the class at:
http://www.as400.ibm.com/db2/db2educ_m.htm
--
Kent Milligan, DB2 & Business Intelligence team
AS/400 Partners In Development
km...@us.removethis.ibm.com GO HAWKEYES!!
(opinions stated are not necessarily those of my employer)