db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sunitha Kambhampati (JIRA)" <derby-...@db.apache.org>
Subject [jira] Commented: (DERBY-384) Doc Review: Tuning Derby
Date Tue, 28 Jun 2005 19:38:04 GMT
    [ http://issues.apache.org/jira/browse/DERBY-384?page=comments#action_12314646 ] 

Sunitha Kambhampati commented on DERBY-384:

some comments on the pdf version, using page numbers on the pdf sheet.

1)     page14:  under performance tradeoffs of large pages: 
Change line  ' By default the database page size is 4096 bytes'  to 
'Derby autotunes the database pagesize, so in case you have long columns, the default pagesize
 for the table is set to 32K else it defaults to 4K'

2)   page16:  Under 'Avoid inserts in autocommit mode if possible' 
There are 2 bullets in this section and the last line  ' Derby will not log the individual
inserts when loading into an empty table using these intefaces'  applies only the second bullet.
 Change the indendation of it to be at the same level as the second bullet else the meaning
is changed. 

3)  page 19 Under the section 'Think about index order'  a few lines below that. 
Remove the line ' Force the index if necessary'  .   SInce there is no way to force the index
if necessary in derby ( no optimizer overrides) 

4) page 19. unders section 'Think about join order', the example and description dont match.
  Remove the example and the description.   We do not have toursDB and mention of it here
is confusing and does not help the user reading the manual.   

5) page 20 and page 21 in my pdf copy have 2 figures which both say Figure 1 . I am assuming
it is because it is a pdf version or something is not right with the linking ?

6) page 23: Section 'Under section Analyzing Statement execution' 
point 4, 
remove the sentence ' Force a better order when necessary'.  ( we dont support optimizer overrides)
add the following : " Another option to check if indexes are being used or not  is to set
derby.language.logQueryPlan=true to print query plans in derby.log. See property derby.language.logQueryPlan.

7)page 34: Section Join Strategies. 
way below on that page.    Change the following line -' The hash table for a hash join is
held in memory and if it gets big enough, it can cause the JVM to run out of memory' to 'The
hash table for a hash join is held in memory and if it gets big enough, it will spill to the
disk '

8) most examples in tuning guide use toursDB examples. There is no mention of this database
and the schema so it does seem a little confusing when it is being mentioned in most places
without  much background on it.  

> Doc Review: Tuning Derby
> ------------------------
>          Key: DERBY-384
>          URL: http://issues.apache.org/jira/browse/DERBY-384
>      Project: Derby
>         Type: Improvement
>   Components: Documentation
>  Environment: all
>     Reporter: Jeff Levitt
>     Priority: Minor
>      Fix For:

> This issue tracks comments for the manual known as Tuning Derby. The deadline for posting
comments is Tuesday, June 28, noon Pacific time.
> Some guidelines to follow when posting comments to this issue are:
> - Try to make clear and concise comments about what you want changed whenever possible.
 Provide concrete comments that say "Please change <original> to <modified>" instead
of generic comments like "This section needs to be rewritten."
> - If you're reviewing the HTML Files copy, include the URL for the page in the review
comment. Obtain the URL like this:
>     * highlight the topic in the left frame
>     * right click
>     * choose "Properties"
>     * copy and paste the address in the pop up box.
> - If you're reviewing the PDF copy, in the review comment:
>     * Include the page number for the PDF, and indicate whether the number is the PDF
sheet number or the printed page number.
>     * Include the title of the section that the problem occurs in. If it's in a subsection,
try to include the hierarchy of titles.
> - Please don't review the HTML Book copy -- it'll be time consuming to match up that
copy with the underlying DITA source.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators:
For more information on JIRA, see:

View raw message