 
| Subject: | |
| From: | |
| Reply To: | |
| Date: | Tue, 6 May 1997 15:49:06 -0600 |
| Content-Type: | text/plain |
| Parts/Attachments: |
|
|
On Tue, 29 Apr 1997 15:09:51 -0600
Keven Miller <[log in to unmask]> wrote:
>An example of what John D. Alleyn-Day mentioned:
> Our ITEM-MASTER was about 90% full at one time and users were beginning to
>get the 2+ minutes wait to insert, ..I mean DBPUT... Checking our output
>from HowMessy we found that the MAX/BLK value was about 500. This means that
>as a cluster of 500 consecutive records exists. Note that this is a MASTER
>set. If DBPUT hashes to the front of this block, and finds the record
taken, >IMAGE serially reads forward until it finds an empty one.
To pick a nit, the 500 MAX/BLK means that there is a cluster of 500 IMAGE
BLOCKS. Depending upon the BlockingFactor of the dataset in question, this
could represent anywhere from 500*BlockingFactor of 1 = 500 IMAGE entries
to 500*BlockingFacotr of 255 = 127,500 IMAGE entries.
When adding IMAGE entries the two major causes for slow performance are
adding entries to a Master dataset with bad clustering as described above
or adding records to a detail dataset with a sorted path which has LONG
chains and the sort key value causes IMAGE to search where to place the
entry within the chain. Both of these conditions have been addressed
within this thread.
>I don't know the availability of HowMessy from ROBELLE, but you can always
try
>something from the CSY. I don't know which, if any, show max/blk, but you can
>check out DBLOADING, DBRUNS, DBSTAT2, there could be others.
/plug
Adager now sends out HOWMESSY with every customer tape shipped. Both the
NM and CM versions can be found in LIBRARY.REGO.
/noplug
Hope this helps,
+---------------+
| |
| r | Ken [log in to unmask]
| e | http://www.adager.com
| g | Ken Paul Tel 208 726-9100
| a | Customer Support Fax 208 726-2822
| d | Adager Corporation
| A | Sun Valley, Idaho 83353-3000 U.S.A.
| |
+---------------+
|
|
|
