You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@commons.apache.org by Paulo Gaspar <pa...@krankikom.de> on 2001/12/23 06:40:31 UTC

Database Connection Pool

With so many reference to database connection pools I have to ask:
 - Are you not happy with DBCP?

Things I miss on DBCP:
 - The wrapping is not complete;
 - Not enough pool behavior control;
 - No easy way of closing all Statements and ResultSets related to a
   Connection.

So, I was NOT happy with DBCP and what I wanted to change was too 
much to make it just by posting patches. So I built one, learning 
from DBCP and using lot of its code in the process.

And I could use some more pairs of hands and eyes to improve it and 
make sure there are no further bugs (it already has some full days 
of testing). Also, I am still not happy with some details and could
use other's opinions.

So I would like to contribute it to the commons. If you want to 
take a look I will post it as I am doing with the logging stuff.

I am asking because it will still take some hours to make sure it
works alone (without the rest of the framework). If you all love
DBCP and do not care to spend time looking at other stuff I can
save that time.

(Anyway, I am starting to divide some of my stuff into subprojects
and I will end up isolating this one too one of these days.)


Differences from DBCP:
  - Complete wrapping, providing the possibility to close all the
    Statements and ResultSets depending of a Connection;

  - Possibility of closing all the Connections (hence Statements
    and ResultSets too) of a "client" DataSource - you can have a
    "server" DataSource that wraps the pool and a "client" 
    DataSource that you can use for a thread (like in a Servlet)
    and close at the end of its use (like at the end of processing
    a Servlet request). This helps to ensure proper cleaning up;

  - More control over the Pool behavior (how it grows and how it 
    shrinks);

  - Different underlying Pool;

  - NO Keyed Pools (my poor brain gets confused with those) and NO
    statement caching (IMHO too much complexity for such low gain);

  - Possibility of using a thread pool for the cleanup stuff.


BTW: it works.

Please tell me if you are curious to see it.


Thanks and have fun,
Paulo Gaspar


> -----Original Message-----
> From: Vincent Massol [mailto:vmassol@octo.com]
> Sent: Saturday, December 22, 2001 8:09 PM
> To: commons-dev@jakarta.apache.org
> Cc: avalon-dev@jakarta.apache.org
> Subject: Synergies between Avalon and Commons projects ?
> 
> 
> I'd like to have your opinion on the possible synergies between Avalon
> and the Commons projects on reuse and sharing.
> 
> Here are some of my thoughts :-)
> 
> * Avalon is a service framework (ok, component and service framework)
> and is suited for lots of projects but some projects may not need/want
> to use the small Avalon wrapper which is layered on top of the java
> classes and may want to use directly the implementation (like using a
> jdbc pool but not as an Avalon component which needs to be managed by a
> component manager)
> 
> * On the other hand, Commons is geared towards providing context neutral
> libraries which could be used in any context.
> 
> * Thus, one solution could be to put in Commons the bare implementation
> (like the jdbc pool implementation) and put in Avalon a component
> wrapper on top of it (and reference the commons jars).
> 
> ...

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Paul Hammant <Pa...@yahoo.com>.
Randy Speh wrote:

>Thanks for the info.  Would you mind telling me the
>differences between Avalon and Fulcrum/Commons.
>
>I just wasn't aware of Avalon for some reason.  I've
>been submerged in Turbine, Torque, Fulcrum and Commons
>code.
>
It is big.  See http://jakarta.apache.org/avalon/

Commons has some overlap with Excalibur only.

Regards,

- Paul H


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Paul Hammant <Pa...@yahoo.com>.
Randy,

>Could you please tell me where the document is on how
>to configure and start and access Avalon's connection
>pool.
>
Wrong list and "using my brain".

Get Avalon (four projects) from CVS.  Use "Find in Files" or equiv to 
find source containing "ConnectionPool" or just "Pool" for more fun.

Regards,

- Paul H


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Berin Loritsch <bl...@apache.org>.
Randy Speh wrote:

> Could you please tell me where the document is on how
> to configure and start and access Avalon's connection
> pool.


The javadocs for the JDBC DataSource are here:

http://jakarta.apache.org/avalon/excalibur/api/org/apache/avalon/excalibur/datasource/JdbcDataSource.html

There is more information in the Avalon Developers Docs in Chapter 3:

http://jakarta.apache.org/avalon/developing/implementing.html

But it is interspersed in instructions for how to use other parts of
Avalon.

It is an Avalon Component, so it is not easily separated.

If you forgo the nice configuration interface, you can set it up programatically
like this:

Class.forName("oracle.jdbc.driver.OracleDriver");
JdbcConnectionFactory factory = public JdbcConnectionFactory("jdbc:oracle:thin:@localhost:1521:ORCL",
                              "scott", "tiger", false /* autocommit? */,
                              "SELECT 1 FROM DUAL",
                              "org.apache.avalon.excalibur.datasource.JdbcConnection");
DefaultPoolController controller = new DefaultPoolController( 2 /* grow amount */ );
JdbcConnectionPool datasourcePool = new JdbcConnectionPool( factory, controller, 2 /* min */, 10 /* max */, false /* autocommit? */

datasourcePool.setTimeout( 500 ); // set time in millis pool will block waiting for a connection to be returned to an empty pool
datasourcePool.initialize();

// the datasourcePool is active now

(Connection) datasourcePool.get();

// when the pool is ready to be shut down
datasourcePool.dispose();


A little explanation follows:

The pool assumes that the driver is actually installed in the classpath, and
the class has been loaded.

The JdbcConnectionFactory is used to create the Connection objects.  The Connection
objects must extend AbstractJdbcConnection.  In Avalon there are two JdbcConnection
implementations: JdbcConnection for JDBC 2.0 compliant drivers, and Jdbc3Connection
for JDBC 3.0 compliant drivers (aka JDK 1.4).

The parameter names for the JdbcConnectionFactory are as follows:

public JdbcConnectionFactory(java.lang.String url,
                              java.lang.String username,
                              java.lang.String password,
                              boolean autoCommit,
                              java.lang.String keepAlive,
                              java.lang.String connectionClass)

The parameter names for the datasourcePool are as follows:

public JdbcConnectionPool(JdbcConnectionFactory factory,
                           DefaultPoolController controller,
                           int min,
                           int max,
                           boolean autoCommit)


The initialization of the pool is asynchronous, and starts when you call initialize();
If you ask for a connection before the pool is fully initialized, it will block until
the pool is ready.  This allows the pool to open new connections (which can take upwards
of a second or more per connection) in the background while your application finishes
initializing.  Most of the time, the pool is done initializing before you request the
first datasource.

There might be a detail or two I am missing, but that is because I use the JdbcDataSource
Component, which takes advantage of Avalon Framework's lifecycle methods.  The full
component takes care of all the details as long as you pass it a valid configuration
object.


> 
> Thank you,
> Randy Speh
> 
> 
> 
> --- Berin Loritsch <bl...@apache.org> wrote:
> 
>>Randy Speh wrote:
>>
>>
>>>Thanks for the info.  Would you mind telling me
>>>
>>the
>>
>>>differences between Avalon and Fulcrum/Commons.
>>>
>>>I just wasn't aware of Avalon for some reason. 
>>>
>>I've
>>
>>>been submerged in Turbine, Torque, Fulcrum and
>>>
>>Commons
>>
>>>code.
>>>
>>
>>Honestly, I can only tell you what Avalon does (I
>>wrote
>>the code for the DB pooling).  I don't know what
>>Fulcrum
>>is much less what it does.  As for Commons, I have
>>only
>>had a cursory look at it, so I am not the one to
>>answer this.
>>
>>
>>
>>>Thanks,
>>>Randy
>>>--- Berin Loritsch <bl...@apache.org> wrote:
>>>
>>>
>>>>Randy Speh wrote:
>>>>
>>>>
>>>>
>>>>>I would be very interested to see what your
>>>>>
>>>>>
>>>>connection
>>>>
>>>>
>>>>>pooling code looks like.  I been trying to choose
>>>>>
>>>>>
>>>>the
>>>>
>>>>
>>>>>most appropriate connection pooling technique and
>>>>>
>>>>>
>>>>have
>>>>
>>>>
>>>>>been considering adapting one myself.  Although,
>>>>>
>>>>>
>>>>I'd
>>>>
>>>>
>>>>>really rather use something from the open source
>>>>>community.
>>>>>
>>>>>
>>>>>
>>>>Currently there is dbcp in Commons, Turbine
>>>>connection pooling,
>>>>and Avalon connection pooling.
>>>>
>>>>Both DBCP and Avalon provide similar semantics
>>>>inspired by
>>>>the official JDBC DataSource specifications--when
>>>>you "close"
>>>>the connection, it returns the Connection to the
>>>>pool.  Turbine
>>>>(if I recall) makes you explicitly return it to
>>>>
>>the
>>
>>>>pool yourself.
>>>>
>>>>I won't get involved with which is the best
>>>>
>>pattern.
>>
>>>>Beyond that basic difference, there is alot of
>>>>difference in the
>>>>manner that the connection pools are configured. 
>>>>Since I don't
>>>>have any real information on DBCP and Turbine
>>>>pooling, I won't
>>>>talk about them.  Here is what Avalon's
>>>>Configuration looks like:
>>>>
>>>><connection>
>>>>  <pool-controller min="2" max="10" grow="2">
>>>>    <keep-alive>SELECT 1 FROM DUAL</keep-alive>
>>>>  </pool-controller>
>>>>
>>>> 
>>>>
>><driver>oracle.jdbc.driver.OracleDriver</driver>
>>
>>>> 
>>>>
>>><dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
>>>
>>>>  <user>scott</user>
>>>>  <password>tiger</password>
>>>></connection>
>>>>
>>>>The Pool Controller specifies how many connections
>>>>the pool starts
>>>>with ("min"), the absolute maximum number of
>>>>connections ("max"),
>>>>and how many new connections it creates at a time
>>>>when it needs
>>>>more ("grow").  It is important to note that the
>>>>maximum is absolute.
>>>>The pool will not go beyond that maximum number. 
>>>>The reason being
>>>>that some drivers have license restrictions and it
>>>>helps the pool
>>>>to not violate the license.  The last part
>>>>("keep-alive") is the
>>>>query used by the pool to determine if the
>>>>connection has been closed
>>>>by the server.  We found out that not all
>>>>
>>databases
>>
>>>>are created
>>>>equal, and while "SELECT 1" works for many
>>>>databases, Oracle requires
>>>>that "1" be selected from the dummy table "DUAL",
>>>>and others like
>>>>Informix require you to select from an
>>>>
>>application's
>>
>>>>table.
>>>>
>>>>You can determine for yourself if this is
>>>>desirable--and possibly
>>>>DBCP might steal ideas to narrow the differences. 
>>>>Who knows....
>>>>
>>>>
>>>>-- 
>>>>
>>>>"They that give up essential liberty to obtain a
>>>>little temporary safety
>>>> deserve neither liberty nor safety."
>>>>                - Benjamin Franklin
>>>>
>>>>
>>>>--
>>>>To unsubscribe, e-mail:  
>>>>
>>><ma...@jakarta.apache.org>
>>>
>>>>For additional commands, e-mail:
>>>><ma...@jakarta.apache.org>
>>>>
>>>
>>>__________________________________________________
>>>Do You Yahoo!?
>>>Send your FREE holiday greetings online!
>>>http://greetings.yahoo.com
>>>
>>>--
>>>To unsubscribe, e-mail:  
>>>
>><ma...@jakarta.apache.org>
>>
>>>For additional commands, e-mail:
>>>
>><ma...@jakarta.apache.org>
>>
>>>.
>>>
>>>
>>>
>>
>>
>>-- 
>>
>>"They that give up essential liberty to obtain a
>>little temporary safety
>>  deserve neither liberty nor safety."
>>                 - Benjamin Franklin
>>
>>
>>--
>>To unsubscribe, e-mail:  
>><ma...@jakarta.apache.org>
>>For additional commands, e-mail:
>><ma...@jakarta.apache.org>
>>
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Send your FREE holiday greetings online!
> http://greetings.yahoo.com
> 
> --
> To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
> 
> .
> 
> 



-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Randy Speh <rw...@yahoo.com>.
Could you please tell me where the document is on how
to configure and start and access Avalon's connection
pool.

Thank you,
Randy Speh



--- Berin Loritsch <bl...@apache.org> wrote:
> Randy Speh wrote:
> 
> > Thanks for the info.  Would you mind telling me
> the
> > differences between Avalon and Fulcrum/Commons.
> > 
> > I just wasn't aware of Avalon for some reason. 
> I've
> > been submerged in Turbine, Torque, Fulcrum and
> Commons
> > code.
> 
> 
> Honestly, I can only tell you what Avalon does (I
> wrote
> the code for the DB pooling).  I don't know what
> Fulcrum
> is much less what it does.  As for Commons, I have
> only
> had a cursory look at it, so I am not the one to
> answer this.
> 
> 
> > 
> > Thanks,
> > Randy
> > --- Berin Loritsch <bl...@apache.org> wrote:
> > 
> >>Randy Speh wrote:
> >>
> >>
> >>>I would be very interested to see what your
> >>>
> >>connection
> >>
> >>>pooling code looks like.  I been trying to choose
> >>>
> >>the
> >>
> >>>most appropriate connection pooling technique and
> >>>
> >>have
> >>
> >>>been considering adapting one myself.  Although,
> >>>
> >>I'd
> >>
> >>>really rather use something from the open source
> >>>community.
> >>>
> >>>
> >>
> >>Currently there is dbcp in Commons, Turbine
> >>connection pooling,
> >>and Avalon connection pooling.
> >>
> >>Both DBCP and Avalon provide similar semantics
> >>inspired by
> >>the official JDBC DataSource specifications--when
> >>you "close"
> >>the connection, it returns the Connection to the
> >>pool.  Turbine
> >>(if I recall) makes you explicitly return it to
> the
> >>pool yourself.
> >>
> >>I won't get involved with which is the best
> pattern.
> >>
> >>Beyond that basic difference, there is alot of
> >>difference in the
> >>manner that the connection pools are configured. 
> >>Since I don't
> >>have any real information on DBCP and Turbine
> >>pooling, I won't
> >>talk about them.  Here is what Avalon's
> >>Configuration looks like:
> >>
> >><connection>
> >>   <pool-controller min="2" max="10" grow="2">
> >>     <keep-alive>SELECT 1 FROM DUAL</keep-alive>
> >>   </pool-controller>
> >>
> >>  
> <driver>oracle.jdbc.driver.OracleDriver</driver>
> >>  
>
>><dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
> >>   <user>scott</user>
> >>   <password>tiger</password>
> >></connection>
> >>
> >>The Pool Controller specifies how many connections
> >>the pool starts
> >>with ("min"), the absolute maximum number of
> >>connections ("max"),
> >>and how many new connections it creates at a time
> >>when it needs
> >>more ("grow").  It is important to note that the
> >>maximum is absolute.
> >>The pool will not go beyond that maximum number. 
> >>The reason being
> >>that some drivers have license restrictions and it
> >>helps the pool
> >>to not violate the license.  The last part
> >>("keep-alive") is the
> >>query used by the pool to determine if the
> >>connection has been closed
> >>by the server.  We found out that not all
> databases
> >>are created
> >>equal, and while "SELECT 1" works for many
> >>databases, Oracle requires
> >>that "1" be selected from the dummy table "DUAL",
> >>and others like
> >>Informix require you to select from an
> application's
> >>table.
> >>
> >>You can determine for yourself if this is
> >>desirable--and possibly
> >>DBCP might steal ideas to narrow the differences. 
> >>Who knows....
> >>
> >>
> >>-- 
> >>
> >>"They that give up essential liberty to obtain a
> >>little temporary safety
> >>  deserve neither liberty nor safety."
> >>                 - Benjamin Franklin
> >>
> >>
> >>--
> >>To unsubscribe, e-mail:  
>
>><ma...@jakarta.apache.org>
> >>For additional commands, e-mail:
> >><ma...@jakarta.apache.org>
> >>
> > 
> > 
> > __________________________________________________
> > Do You Yahoo!?
> > Send your FREE holiday greetings online!
> > http://greetings.yahoo.com
> > 
> > --
> > To unsubscribe, e-mail:  
> <ma...@jakarta.apache.org>
> > For additional commands, e-mail:
> <ma...@jakarta.apache.org>
> > 
> > .
> > 
> > 
> 
> 
> 
> -- 
> 
> "They that give up essential liberty to obtain a
> little temporary safety
>   deserve neither liberty nor safety."
>                  - Benjamin Franklin
> 
> 
> --
> To unsubscribe, e-mail:  
> <ma...@jakarta.apache.org>
> For additional commands, e-mail:
> <ma...@jakarta.apache.org>
> 


__________________________________________________
Do You Yahoo!?
Send your FREE holiday greetings online!
http://greetings.yahoo.com

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Berin Loritsch <bl...@apache.org>.
Randy Speh wrote:

> Thanks for the info.  Would you mind telling me the
> differences between Avalon and Fulcrum/Commons.
> 
> I just wasn't aware of Avalon for some reason.  I've
> been submerged in Turbine, Torque, Fulcrum and Commons
> code.


Honestly, I can only tell you what Avalon does (I wrote
the code for the DB pooling).  I don't know what Fulcrum
is much less what it does.  As for Commons, I have only
had a cursory look at it, so I am not the one to answer this.


> 
> Thanks,
> Randy
> --- Berin Loritsch <bl...@apache.org> wrote:
> 
>>Randy Speh wrote:
>>
>>
>>>I would be very interested to see what your
>>>
>>connection
>>
>>>pooling code looks like.  I been trying to choose
>>>
>>the
>>
>>>most appropriate connection pooling technique and
>>>
>>have
>>
>>>been considering adapting one myself.  Although,
>>>
>>I'd
>>
>>>really rather use something from the open source
>>>community.
>>>
>>>
>>
>>Currently there is dbcp in Commons, Turbine
>>connection pooling,
>>and Avalon connection pooling.
>>
>>Both DBCP and Avalon provide similar semantics
>>inspired by
>>the official JDBC DataSource specifications--when
>>you "close"
>>the connection, it returns the Connection to the
>>pool.  Turbine
>>(if I recall) makes you explicitly return it to the
>>pool yourself.
>>
>>I won't get involved with which is the best pattern.
>>
>>Beyond that basic difference, there is alot of
>>difference in the
>>manner that the connection pools are configured. 
>>Since I don't
>>have any real information on DBCP and Turbine
>>pooling, I won't
>>talk about them.  Here is what Avalon's
>>Configuration looks like:
>>
>><connection>
>>   <pool-controller min="2" max="10" grow="2">
>>     <keep-alive>SELECT 1 FROM DUAL</keep-alive>
>>   </pool-controller>
>>
>>   <driver>oracle.jdbc.driver.OracleDriver</driver>
>>  
>><dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
>>   <user>scott</user>
>>   <password>tiger</password>
>></connection>
>>
>>The Pool Controller specifies how many connections
>>the pool starts
>>with ("min"), the absolute maximum number of
>>connections ("max"),
>>and how many new connections it creates at a time
>>when it needs
>>more ("grow").  It is important to note that the
>>maximum is absolute.
>>The pool will not go beyond that maximum number. 
>>The reason being
>>that some drivers have license restrictions and it
>>helps the pool
>>to not violate the license.  The last part
>>("keep-alive") is the
>>query used by the pool to determine if the
>>connection has been closed
>>by the server.  We found out that not all databases
>>are created
>>equal, and while "SELECT 1" works for many
>>databases, Oracle requires
>>that "1" be selected from the dummy table "DUAL",
>>and others like
>>Informix require you to select from an application's
>>table.
>>
>>You can determine for yourself if this is
>>desirable--and possibly
>>DBCP might steal ideas to narrow the differences. 
>>Who knows....
>>
>>
>>-- 
>>
>>"They that give up essential liberty to obtain a
>>little temporary safety
>>  deserve neither liberty nor safety."
>>                 - Benjamin Franklin
>>
>>
>>--
>>To unsubscribe, e-mail:  
>><ma...@jakarta.apache.org>
>>For additional commands, e-mail:
>><ma...@jakarta.apache.org>
>>
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Send your FREE holiday greetings online!
> http://greetings.yahoo.com
> 
> --
> To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
> 
> .
> 
> 



-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


RE: Database Connection Pool

Posted by Paulo Gaspar <pa...@krankikom.de>.
My code is a mix of DBCP with a bit of learning from 
PoolMan.

Neither Avalon nor DBCP wrap all Statement and ResultSet
interfaces and make sure it is impossible for them to be
used after their Connection is returned to the pool. 
Avalon does nothing in that direction and DBCP is in 
this respect still a bit incomplete (or was when I 
forked).

This allows incorrect use to mess with the pool and the 
behavior of the Connections. It also makes it hard to
implement a timeout policy for active Connection's, 
which sometimes is quite useful.


Besides, I missed a bit more of control over the pace
and the way the Pool grows and shrinks.

(You only need to wrap the ResultSet for JDBC drivers 
incorrectly implemented. Closing the Statement should 
close the ResultSet.)

As in Avalon and in DBCP, there is a separate generic 
pool that is used to implement the database pool. It
also uses the standard DataSource and Connection 
interfaces.

There is neither Statement nor ResultSet caching. In
my experience ( <= read this as "with the stuff I do")
I find no big performance gain in caching Statements
and I believe that ResultSets should be cached 
elsewhere.


Anyway, if you are looking for the ideal Database Pool 
implementations you should also take a look at PoolMan:
  http://www.codestudio.com/

It is LGPL and it is not perfect, but even when this 
is an obstacle you can always learn a lot from it.


Have fun,
Paulo Gaspar

http://www.krankikom.de
http://www.ruhronline.de
 

> -----Original Message-----
> From: Randy Speh [mailto:rwspeh@yahoo.com]
> Sent: Thursday, December 27, 2001 8:14 PM
> 
> 
> Thanks for the info.  Would you mind telling me the
> differences between Avalon and Fulcrum/Commons.
> 
> I just wasn't aware of Avalon for some reason.  I've
> been submerged in Turbine, Torque, Fulcrum and Commons
> code.
> 
> Thanks,
> Randy
> --- Berin Loritsch <bl...@apache.org> wrote:
> > Randy Speh wrote:
> > 
> > > I would be very interested to see what your
> > connection
> > > pooling code looks like.  I been trying to choose
> > the
> > > most appropriate connection pooling technique and
> > have
> > > been considering adapting one myself.  Although,
> > I'd
> > > really rather use something from the open source
> > > community.
> > > 
> > 
> > 
> > Currently there is dbcp in Commons, Turbine
> > connection pooling,
> > and Avalon connection pooling.
> > 
> > Both DBCP and Avalon provide similar semantics
> > inspired by
> > the official JDBC DataSource specifications--when
> > you "close"
> > the connection, it returns the Connection to the
> > pool.  Turbine
> > (if I recall) makes you explicitly return it to the
> > pool yourself.
> > 
> > I won't get involved with which is the best pattern.
> > 
> > Beyond that basic difference, there is alot of
> > difference in the
> > manner that the connection pools are configured. 
> > Since I don't
> > have any real information on DBCP and Turbine
> > pooling, I won't
> > talk about them.  Here is what Avalon's
> > Configuration looks like:
> > 
> > <connection>
> >    <pool-controller min="2" max="10" grow="2">
> >      <keep-alive>SELECT 1 FROM DUAL</keep-alive>
> >    </pool-controller>
> > 
> >    <driver>oracle.jdbc.driver.OracleDriver</driver>
> >   
> > <dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
> >    <user>scott</user>
> >    <password>tiger</password>
> > </connection>
> > 
> > The Pool Controller specifies how many connections
> > the pool starts
> > with ("min"), the absolute maximum number of
> > connections ("max"),
> > and how many new connections it creates at a time
> > when it needs
> > more ("grow").  It is important to note that the
> > maximum is absolute.
> > The pool will not go beyond that maximum number. 
> > The reason being
> > that some drivers have license restrictions and it
> > helps the pool
> > to not violate the license.  The last part
> > ("keep-alive") is the
> > query used by the pool to determine if the
> > connection has been closed
> > by the server.  We found out that not all databases
> > are created
> > equal, and while "SELECT 1" works for many
> > databases, Oracle requires
> > that "1" be selected from the dummy table "DUAL",
> > and others like
> > Informix require you to select from an application's
> > table.
> > 
> > You can determine for yourself if this is
> > desirable--and possibly
> > DBCP might steal ideas to narrow the differences. 
> > Who knows....
> > 
> > 
> > -- 
> > 
> > "They that give up essential liberty to obtain a
> > little temporary safety
> >   deserve neither liberty nor safety."
> >                  - Benjamin Franklin
> > 
> > 
> > --
> > To unsubscribe, e-mail:  
> > <ma...@jakarta.apache.org>
> > For additional commands, e-mail:
> > <ma...@jakarta.apache.org>
> > 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Send your FREE holiday greetings online!
> http://greetings.yahoo.com
> 
> --
> To unsubscribe, e-mail:   
> <ma...@jakarta.apache.org>
> For additional commands, e-mail: 
> <ma...@jakarta.apache.org>
> 

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Randy Speh <rw...@yahoo.com>.
Thanks for the info.  Would you mind telling me the
differences between Avalon and Fulcrum/Commons.

I just wasn't aware of Avalon for some reason.  I've
been submerged in Turbine, Torque, Fulcrum and Commons
code.

Thanks,
Randy
--- Berin Loritsch <bl...@apache.org> wrote:
> Randy Speh wrote:
> 
> > I would be very interested to see what your
> connection
> > pooling code looks like.  I been trying to choose
> the
> > most appropriate connection pooling technique and
> have
> > been considering adapting one myself.  Although,
> I'd
> > really rather use something from the open source
> > community.
> > 
> 
> 
> Currently there is dbcp in Commons, Turbine
> connection pooling,
> and Avalon connection pooling.
> 
> Both DBCP and Avalon provide similar semantics
> inspired by
> the official JDBC DataSource specifications--when
> you "close"
> the connection, it returns the Connection to the
> pool.  Turbine
> (if I recall) makes you explicitly return it to the
> pool yourself.
> 
> I won't get involved with which is the best pattern.
> 
> Beyond that basic difference, there is alot of
> difference in the
> manner that the connection pools are configured. 
> Since I don't
> have any real information on DBCP and Turbine
> pooling, I won't
> talk about them.  Here is what Avalon's
> Configuration looks like:
> 
> <connection>
>    <pool-controller min="2" max="10" grow="2">
>      <keep-alive>SELECT 1 FROM DUAL</keep-alive>
>    </pool-controller>
> 
>    <driver>oracle.jdbc.driver.OracleDriver</driver>
>   
> <dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
>    <user>scott</user>
>    <password>tiger</password>
> </connection>
> 
> The Pool Controller specifies how many connections
> the pool starts
> with ("min"), the absolute maximum number of
> connections ("max"),
> and how many new connections it creates at a time
> when it needs
> more ("grow").  It is important to note that the
> maximum is absolute.
> The pool will not go beyond that maximum number. 
> The reason being
> that some drivers have license restrictions and it
> helps the pool
> to not violate the license.  The last part
> ("keep-alive") is the
> query used by the pool to determine if the
> connection has been closed
> by the server.  We found out that not all databases
> are created
> equal, and while "SELECT 1" works for many
> databases, Oracle requires
> that "1" be selected from the dummy table "DUAL",
> and others like
> Informix require you to select from an application's
> table.
> 
> You can determine for yourself if this is
> desirable--and possibly
> DBCP might steal ideas to narrow the differences. 
> Who knows....
> 
> 
> -- 
> 
> "They that give up essential liberty to obtain a
> little temporary safety
>   deserve neither liberty nor safety."
>                  - Benjamin Franklin
> 
> 
> --
> To unsubscribe, e-mail:  
> <ma...@jakarta.apache.org>
> For additional commands, e-mail:
> <ma...@jakarta.apache.org>
> 


__________________________________________________
Do You Yahoo!?
Send your FREE holiday greetings online!
http://greetings.yahoo.com

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Berin Loritsch <bl...@apache.org>.
Randy Speh wrote:

> I would be very interested to see what your connection
> pooling code looks like.  I been trying to choose the
> most appropriate connection pooling technique and have
> been considering adapting one myself.  Although, I'd
> really rather use something from the open source
> community.
> 


Currently there is dbcp in Commons, Turbine connection pooling,
and Avalon connection pooling.

Both DBCP and Avalon provide similar semantics inspired by
the official JDBC DataSource specifications--when you "close"
the connection, it returns the Connection to the pool.  Turbine
(if I recall) makes you explicitly return it to the pool yourself.

I won't get involved with which is the best pattern.

Beyond that basic difference, there is alot of difference in the
manner that the connection pools are configured.  Since I don't
have any real information on DBCP and Turbine pooling, I won't
talk about them.  Here is what Avalon's Configuration looks like:

<connection>
   <pool-controller min="2" max="10" grow="2">
     <keep-alive>SELECT 1 FROM DUAL</keep-alive>
   </pool-controller>

   <driver>oracle.jdbc.driver.OracleDriver</driver>
   <dburl>jdbc:oracle:thin:@localhost:1521:ORCL</dburl>
   <user>scott</user>
   <password>tiger</password>
</connection>

The Pool Controller specifies how many connections the pool starts
with ("min"), the absolute maximum number of connections ("max"),
and how many new connections it creates at a time when it needs
more ("grow").  It is important to note that the maximum is absolute.
The pool will not go beyond that maximum number.  The reason being
that some drivers have license restrictions and it helps the pool
to not violate the license.  The last part ("keep-alive") is the
query used by the pool to determine if the connection has been closed
by the server.  We found out that not all databases are created
equal, and while "SELECT 1" works for many databases, Oracle requires
that "1" be selected from the dummy table "DUAL", and others like
Informix require you to select from an application's table.

You can determine for yourself if this is desirable--and possibly
DBCP might steal ideas to narrow the differences.  Who knows....


-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Database Connection Pool

Posted by Randy Speh <rw...@yahoo.com>.
I would be very interested to see what your connection
pooling code looks like.  I been trying to choose the
most appropriate connection pooling technique and have
been considering adapting one myself.  Although, I'd
really rather use something from the open source
community.

Randy Speh

--- Paulo Gaspar <pa...@krankikom.de> wrote:
> With so many reference to database connection pools
> I have to ask:
>  - Are you not happy with DBCP?
> 
> Things I miss on DBCP:
>  - The wrapping is not complete;
>  - Not enough pool behavior control;
>  - No easy way of closing all Statements and
> ResultSets related to a
>    Connection.
> 
> So, I was NOT happy with DBCP and what I wanted to
> change was too 
> much to make it just by posting patches. So I built
> one, learning 
> from DBCP and using lot of its code in the process.
> 
> And I could use some more pairs of hands and eyes to
> improve it and 
> make sure there are no further bugs (it already has
> some full days 
> of testing). Also, I am still not happy with some
> details and could
> use other's opinions.
> 
> So I would like to contribute it to the commons. If
> you want to 
> take a look I will post it as I am doing with the
> logging stuff.
> 
> I am asking because it will still take some hours to
> make sure it
> works alone (without the rest of the framework). If
> you all love
> DBCP and do not care to spend time looking at other
> stuff I can
> save that time.
> 
> (Anyway, I am starting to divide some of my stuff
> into subprojects
> and I will end up isolating this one too one of
> these days.)
> 
> 
> Differences from DBCP:
>   - Complete wrapping, providing the possibility to
> close all the
>     Statements and ResultSets depending of a
> Connection;
> 
>   - Possibility of closing all the Connections
> (hence Statements
>     and ResultSets too) of a "client" DataSource -
> you can have a
>     "server" DataSource that wraps the pool and a
> "client" 
>     DataSource that you can use for a thread (like
> in a Servlet)
>     and close at the end of its use (like at the end
> of processing
>     a Servlet request). This helps to ensure proper
> cleaning up;
> 
>   - More control over the Pool behavior (how it
> grows and how it 
>     shrinks);
> 
>   - Different underlying Pool;
> 
>   - NO Keyed Pools (my poor brain gets confused with
> those) and NO
>     statement caching (IMHO too much complexity for
> such low gain);
> 
>   - Possibility of using a thread pool for the
> cleanup stuff.
> 
> 
> BTW: it works.
> 
> Please tell me if you are curious to see it.
> 
> 
> Thanks and have fun,
> Paulo Gaspar
> 
> 
> > -----Original Message-----
> > From: Vincent Massol [mailto:vmassol@octo.com]
> > Sent: Saturday, December 22, 2001 8:09 PM
> > To: commons-dev@jakarta.apache.org
> > Cc: avalon-dev@jakarta.apache.org
> > Subject: Synergies between Avalon and Commons
> projects ?
> > 
> > 
> > I'd like to have your opinion on the possible
> synergies between Avalon
> > and the Commons projects on reuse and sharing.
> > 
> > Here are some of my thoughts :-)
> > 
> > * Avalon is a service framework (ok, component and
> service framework)
> > and is suited for lots of projects but some
> projects may not need/want
> > to use the small Avalon wrapper which is layered
> on top of the java
> > classes and may want to use directly the
> implementation (like using a
> > jdbc pool but not as an Avalon component which
> needs to be managed by a
> > component manager)
> > 
> > * On the other hand, Commons is geared towards
> providing context neutral
> > libraries which could be used in any context.
> > 
> > * Thus, one solution could be to put in Commons
> the bare implementation
> > (like the jdbc pool implementation) and put in
> Avalon a component
> > wrapper on top of it (and reference the commons
> jars).
> > 
> > ...
> 
> --
> To unsubscribe, e-mail:  
> <ma...@jakarta.apache.org>
> For additional commands, e-mail:
> <ma...@jakarta.apache.org>
> 


__________________________________________________
Do You Yahoo!?
Send your FREE holiday greetings online!
http://greetings.yahoo.com

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>