You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user-java@ibatis.apache.org by Clinton Begin <cl...@gmail.com> on 2006/03/01 02:52:09 UTC
Re: performnace of iBATIS
Thanks you for your honesty and the clarification.
Cheers,
Clinton
On 2/28/06, Tony Qian <da...@aol.com> wrote:
>
> Brandon and all,
>
> I convinced myself now that iBATIS is almost as good as JDBC for simple
> query. I made mistake that I forgot to turn off ibatis debug log. After
> I turned it off, iBATIS and JDBC call were very close.
>
> Sorry for confusion and wish the best for iBATIS and this community.
>
> Thanks,
> Tony
>
> Brandon Goodin wrote on 2/28/2006, 11:15 AM:
>
> > Please post all the code you use to test jdbc and ibatis. There are a
> > few mistakes that people can often make. One is that you don't perform
> > the resultset to object translation. The next is that the test could
> > be loading the sqlmap each time it is run. iBATIS will have initial
> > overhead because it is parsing and loading the sqlmaps into memory.
> >
> > Brandon
> >
> > On 2/28/06, Tony Qian <da...@aol.com> wrote:
> > > Clinton and all,
> > >
> > > Let me first thank you guys for quick response. I believe we have
> > the best
> > > tech support among open source community, either from iBATIS team or
> > from
> > > iBATIS users.
> > >
> > > I used a very simple query which fetches all records in the table (72
> > > rows). Here are some info for my test.
> > >
> > > database: MySql 5.0, query cache is turned on.
> > > iBATIS setting:
> > >
> > > <settings cacheModelsEnabled="true" // but no cache is set
> > for query
> > > enhancementEnabled="true"
> > > lazyLoadingEnabled="false"
> > > maxRequests="32"
> > > maxSessions="10"
> > > maxTransactions="5"
> > > useStatementNamespaces="false" />
> > > <property name="Pool.PingEnabled" value="false"/>
> > >
> > > xml mapping for query:
> > > <typeAlias alias="sessionParameter"
> > >
> >
> type="com.netscape.isp.business.management.session.SimpleSessionParameter
> "/>
>
> >
> > > <select id="getSessionParameterList"
> > resultClass="sessionParameter" >
> > > SELECT
> > > parameter_id as parameterId,
> > > parameter_name as name,
> > > description as description
> > > FROM SESSION_PARAMETER ORDER by parameter_name
> > > </select>
> > >
> > > Java code:
> > > try{
> > > long stime = System.currentTimeMillis();
> > > Context ctx = new InitialContext();
> > > DataSource dataSource = (DataSource)
> > > ctx.lookup("java:comp/env/jdbc/quickstart");
> > > Connection connection = dataSource.getConnection();
> > > Statement stmt = connection.createStatement();
> > > String sql = "SELECT parameter_id, parameter_name,
> > description
> > > FROM SESSION_PARAMETER;" ;
> > > stmt.execute(sql);
> > > ResultSet rS = stmt.getResultSet();
> > > List list = new ArrayList();
> > > if (rS.first()) {
> > > do {
> > > list.add(new SimpleSessionParameter(
> > > rS.getString("parameter_id"),
> > > rS.getString("parameter_name"),
> > > rS.getString("description")));
> > > } while (rS.next());
> > > } else {
> > > throw new NotFoundException("no records were found");
> > > }
> > > rS.close();
> > > stmt.close();
> > > out.println("executing time is " +
> > (System.currentTimeMillis()
> > > - stime) +
> > > " current time =" +System.currentTimeMillis() + "
> > > startTime = " + stime);
> > > stime = System.currentTimeMillis();
> > > List parameter = new SimpleSessionParameterList(); //
> > > basically it calls "list =
> > sqlMap.queryForList("getSessionParameterList",
> > > null);"
> > > out.println("Ibatis executing time is " +
> > > (System.currentTimeMillis() - stime) +
> > > " current time ="
> > > +System.currentTimeMillis() + "startTime = " + stime);
> > > }catch (Exception e){
> > > out.println(" get exception" + e.getMessage());
> > > }
> > >
> > >
> > > Test results:
> > > JDBC executing time is 30 current time =1141136720755 startTime =
> > > 1141136720725
> > > Ibatis executing time is 100 current time =1141136720855startTime =
> > > 1141136720755
> > >
> > > JDBC executing time is 40 current time =1141136723448 startTime =
> > > 1141136723408
> > > Ibatis executing time is 110 current time =1141136723558startTime =
> > > 1141136723448
> > >
> > > JDBC executing time is 20 current time =1141136725741 startTime =
> > > 1141136725721
> > > Ibatis executing time is 100 current time =1141136725841startTime =
> > > 1141136725741
> > >
> > > executing time is 30 current time =1141137529882 startTime =
> > 1141137529852
> > > Ibatis executing time is 101 current time =1141137529983startTime =
> > > 1141137529882
> > >
> > > I appreciate your help.
> > > Tony
> > >
> > > Clinton Begin wrote on 2/28/2006, 1:59 AM:
> > >
> > >
> > > Post your test code, and I'll show you what's wrong with it.
> > >
> > > Clinton
> > >
> > >
> > >
> > > On 2/27/06, DaqiQian2@aol.com <Da...@aol.com> wrote:
> > >
> > > Sven,
> > >
> > > thanks for your response. I did some preliminary test on iBATIS'
> > > performance. For JDBC (MySql 5.0), i recorded time from establishing
> > > connection, result set, and mapping the result to objects. For list of
> > > simple objects, it seems to me that iBATIS (no caching and
> > lazyloading) is
> > > 3-5 times slower than using JDBC. Of course, i believe my xml
> > mapping has
> > > room to improve.
> > >
> > > The reason I asked is that I need some stats to persuade myself and
> > > coworkers to accept iBATIS as a data persistence tool for our
> > relatively
> > > heavily loaded servers.
> > >
> > > btw, we just pushed a project using iBATIS to QE. For that project,
> > > performance is not big issue.
> > >
> > > Thanks,
> > > Tony
> > >
> > >
> >
>
>
>
Re: performnace of iBATIS
Posted by Ben Munat <be...@munat.com>.
/Warning: rambling aside ahead:/ I remember the eye-opening I had in OS theory class when
we talked about IO-bound vs. processor-bound... and the fact that going to disk is often a
*thousand times slower* than getting something out of ram (microseconds vs. nanoseconds).
So, if the database access winds up basically being the same between JDBC and ibatis, then
I would be very surprised if there could be that much difference between the two overall
times.
It's like comparing the distance from the ground to the top of two people's head, but with
them both standing on top of a skyscraper.... once you bring the skyscraper into the
equation, we all look to be about the same height!
b
Clinton Begin wrote:
>
> Thanks you for your honesty and the clarification.
>
> Cheers,
> Clinton
>
>
> On 2/28/06, *Tony Qian* <daqiqian2@aol.com <ma...@aol.com>>
> wrote:
>
> Brandon and all,
>
> I convinced myself now that iBATIS is almost as good as JDBC for simple
> query. I made mistake that I forgot to turn off ibatis debug log. After
> I turned it off, iBATIS and JDBC call were very close.
>
> Sorry for confusion and wish the best for iBATIS and this community.
>
> Thanks,
> Tony
>
> Brandon Goodin wrote on 2/28/2006, 11:15 AM:
>
> > Please post all the code you use to test jdbc and ibatis. There are a
> > few mistakes that people can often make. One is that you don't
> perform
> > the resultset to object translation. The next is that the test could
> > be loading the sqlmap each time it is run. iBATIS will have initial
> > overhead because it is parsing and loading the sqlmaps into memory.
> >
> > Brandon
> >
> > On 2/28/06, Tony Qian <daqiqian2@aol.com
> <ma...@aol.com>> wrote:
> > > Clinton and all,
> > >
> > > Let me first thank you guys for quick response. I believe we have
> > the best
> > > tech support among open source community, either from iBATIS
> team or
> > from
> > > iBATIS users.
> > >
> > > I used a very simple query which fetches all records in the
> table (72
> > > rows). Here are some info for my test.
> > >
> > > database: MySql 5.0, query cache is turned on.
> > > iBATIS setting:
> > >
> > > <settings cacheModelsEnabled="true" // but no cache is set
> > for query
> > > enhancementEnabled="true"
> > > lazyLoadingEnabled="false"
> > > maxRequests="32"
> > > maxSessions="10"
> > > maxTransactions="5"
> > > useStatementNamespaces="false" />
> > > <property name="Pool.PingEnabled" value="false"/>
> > >
> > > xml mapping for query:
> > > <typeAlias alias="sessionParameter"
> > >
> >
> type="com.netscape.isp.business.management.session.SimpleSessionParameter"/>
>
> >
> > > <select id="getSessionParameterList"
> > resultClass="sessionParameter" >
> > > SELECT
> > > parameter_id as parameterId,
> > > parameter_name as name,
> > > description as description
> > > FROM SESSION_PARAMETER ORDER by parameter_name
> > > </select>
> > >
> > > Java code:
> > > try{
> > > long stime = System.currentTimeMillis ();
> > > Context ctx = new InitialContext();
> > > DataSource dataSource = (DataSource)
> > > ctx.lookup("java:comp/env/jdbc/quickstart");
> > > Connection connection = dataSource.getConnection();
> > > Statement stmt = connection.createStatement();
> > > String sql = "SELECT parameter_id, parameter_name,
> > description
> > > FROM SESSION_PARAMETER;" ;
> > > stmt.execute(sql);
> > > ResultSet rS = stmt.getResultSet();
> > > List list = new ArrayList();
> > > if (rS.first()) {
> > > do {
> > > list.add(new SimpleSessionParameter(
> > > rS.getString("parameter_id"),
> > > rS.getString("parameter_name"),
> > > rS.getString("description")));
> > > } while (rS.next());
> > > } else {
> > > throw new NotFoundException("no records were
> found");
> > > }
> > > rS.close();
> > > stmt.close();
> > > out.println("executing time is " +
> > (System.currentTimeMillis()
> > > - stime) +
> > > " current time ="
> +System.currentTimeMillis() + "
> > > startTime = " + stime);
> > > stime = System.currentTimeMillis ();
> > > List parameter = new
> SimpleSessionParameterList(); //
> > > basically it calls "list =
> > sqlMap.queryForList("getSessionParameterList",
> > > null);"
> > > out.println("Ibatis executing time is " +
> > > (System.currentTimeMillis() - stime) +
> > > " current time ="
> > > +System.currentTimeMillis() + "startTime = " + stime);
> > > }catch (Exception e){
> > > out.println(" get exception" + e.getMessage());
> > > }
> > >
> > >
> > > Test results:
> > > JDBC executing time is 30 current time =1141136720755 startTime =
> > > 1141136720725
> > > Ibatis executing time is 100 current time
> =1141136720855startTime =
> > > 1141136720755
> > >
> > > JDBC executing time is 40 current time =1141136723448 startTime =
> > > 1141136723408
> > > Ibatis executing time is 110 current time
> =1141136723558startTime =
> > > 1141136723448
> > >
> > > JDBC executing time is 20 current time =1141136725741 startTime =
> > > 1141136725721
> > > Ibatis executing time is 100 current time
> =1141136725841startTime =
> > > 1141136725741
> > >
> > > executing time is 30 current time =1141137529882 startTime =
> > 1141137529852
> > > Ibatis executing time is 101 current time
> =1141137529983startTime =
> > > 1141137529882
> > >
> > > I appreciate your help.
> > > Tony
> > >
> > > Clinton Begin wrote on 2/28/2006, 1:59 AM:
> > >
> > >
> > > Post your test code, and I'll show you what's wrong with it.
> > >
> > > Clinton
> > >
> > >
> > >
> > > On 2/27/06, DaqiQian2@aol.com <ma...@aol.com>
> <DaqiQian2@aol.com <ma...@aol.com>> wrote:
> > >
> > > Sven,
> > >
> > > thanks for your response. I did some preliminary test on iBATIS'
> > > performance. For JDBC (MySql 5.0), i recorded time from
> establishing
> > > connection, result set, and mapping the result to objects. For
> list of
> > > simple objects, it seems to me that iBATIS (no caching and
> > lazyloading) is
> > > 3-5 times slower than using JDBC. Of course, i believe my xml
> > mapping has
> > > room to improve.
> > >
> > > The reason I asked is that I need some stats to persuade
> myself and
> > > coworkers to accept iBATIS as a data persistence tool for our
> > relatively
> > > heavily loaded servers.
> > >
> > > btw, we just pushed a project using iBATIS to QE. For that
> project,
> > > performance is not big issue.
> > >
> > > Thanks,
> > > Tony
> > >
> > >
> >
>
>
>