You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by Eranda Sooriyabandara <07...@gmail.com> on 2009/04/01 14:26:39 UTC

Re: [GSoC] derby-optimizerFixes

Hi ,
I have a godackground in Derby Server and just  i sarch about optimizer work
and as a new to the open source project i wish to work on the project Derby
testandfix as you mentioned. Also I feel like that it is hard to follow(not
much) up a project like that under my current knoladge.So I am getting
familer with Derby community and the derby testing . Thanks
Eranda

Re: [GSoC] derby-optimizerFixes

Posted by Bryan Pendleton <bp...@amberpoint.com>.
Eranda Sooriyabandara wrote:
> Hi Bryan,
> About DERBY-3892: The problem i faced while testing is because, in 
> Netbeans i did not add the jdbc library before testing it.I used the 
> attachment you gave Main_Updated.java .So now i got good results. I like 
> to show you some of my results i got during my analysis,

Hi Eranda,

Thanks for investigating this problem, and I'm glad you've got your
environment issues straightened out.

I'm not sure exactly which version of the Main_Updated.java test
program you are using. This is important, because some of the
versions demonstrate the problem, and some do not.

You might want to fetch *all* the versions of the test program to
your machine, and diff them against each other, and then match those
versions up to the comments in the JIRA issue, and see if the
behavior makes more sense to you then.

thanks,

bryan


Re: [GSoC] derby-optimizerFixes

Posted by Bryan Pendleton <bp...@amberpoint.com>.
Eranda Sooriyabandara wrote:
> Hi Bryan,
> Now what are the thing we can do other than Learning about the Subject matter?

Hi Eranda,

Regarding your GSoC application, I believe that you can continue to update
your application during the application process. On the Apache side, the
evaluation committee is looking for things like: involvement in the community,
and committment to the open source process:  http://wiki.apache.org/general/RankingProcess
On the Google side, I think they are looking for you to explain how
this will help you improve your development skills:
http://socghop.appspot.com/document/show/program/google/gsoc2009/faqs#student_app

You could take a look at a few of the other Derby issues. I think that
Derby-3892 is a somewhat complex issue, there might be others that
are simpler. Try looking at some of the ones on this report:
https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&pid=10594&sorter/order=DESC&sorter/field=priority&resolution=-1&component=12310640
Find one that interests you, assign it to yourself, and try to fix it.

A few other things that would be great:

  - Have you verified that you can successfully run the regression
    tests for Derby in your environment? You should get 100% pass with
    the current trunk.

  - The Derby 10.5 release is almost ready, and we're doing some final
    testing. You could download the release and help us get it ready,
    for example by finding one of the new features and doing some
    testing on it: http://wiki.apache.org/db-derby/TenFiveOneBuddyTesting

  - You can look at Derby issues which are marked "patch-available" and
    help review the patches by studying them, making comments, and downloading
    and applying and testing them: http://wiki.apache.org/db-derby/ForNewDevelopers

I know these things take time; I hope you are able to find some time to
get involved and stay involved in Derby, as I think it will be a rewarding
process for you.

thanks,

bryan






Re: [GSoC] derby-optimizerFixes

Posted by Eranda Sooriyabandara <07...@gmail.com>.
Hi Bryan,
About DERBY-3892: The problem i faced while testing is because, in Netbeans
i did not add the jdbc library before testing it.I used the attachment you
gave Main_Updated.java .So now i got good results. I like to show you some
of my results i got during my analysis,
  Rows         Insertion time(s)      Querying time(s)
     100             2.281                       2.346
    1000            3.31                         3.388
    1500            4.497                       4.575
    2000            7.717                       7.807
    5000          16.064                     16.143
  10000           23.711                     23.797
  20000           37.76                       37.926
  50000           91.748                     92.086
100000          185.492                   185.98

I check  the 2  two insertion separately (Messaged User,Values )
When only data inserted to Messaged_user(insert to values table removed from
the code) table only it took 80.926s to insertion and 80.992s for querying.
When only data inserted to Values(insert to Messaged_User table removed from
the code) table only it took 77.054s to insertion and 77.16s for querying.
sum of the two in separate insertions=157.98s
sum of the two in separate querying=157.98s

Thats 30s different from what i got when the two insertions are in same
coding.
I try another time with two insertion at the same time.  Amazingly i got
141.721s for insertion and 142.21s for querying. Try for another gave
161.523s for insertion and 162.014s for querying.
So i decided that bug must be on the "insertion"
Next i include the Statement class for execution inside the loop
Insted of ,
PreparedStatement ps = con.prepareStatement
                ("INSERT INTO Messaged_Users VALUES (?, ?, DEFAULT,
DEFAULT)");
            for (i=0; i<100000; i++) {
                ps.setString(1, ""+i);
                ps.setString(2, "User" + i);
                ps.executeUpdate();
            }
I used,
Statement stmt=con.createStatement();
            for (i=0; i<100000; i++) {
                stmt.executeUpdate("INSERT INTO Users VALUES ('"+i+"',
'Eranda')");
            }
as a test,  But i waited 15s for over the insertion and querying and stopped
the process.

Can you show me a path that i can go on further? Thanks
Eranda