You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Philip Williams (Created) (JIRA)" <ji...@apache.org> on 2011/10/11 00:48:30 UTC

[jira] [Created] (DBCP-368) determine which connections to hold in pool by relative value

determine which connections to hold in pool by relative value
-------------------------------------------------------------

                 Key: DBCP-368
                 URL: https://issues.apache.org/jira/browse/DBCP-368
             Project: Commons Dbcp
          Issue Type: New Feature
            Reporter: Philip Williams


Situation: a Firebird-Classic database (each connection has independent metadata cache), in a metadata-heavy environment; many "quick" actions will never force the database server to load full metadata, but a few will, in our case resulting in 200MB of metadata cache in memory (takes about 3 seconds to load, visible latency to the user); any connection that has done this is "more valuable" to us, and we'd prefer to keep those connections in the pool, and remove idle connections which haven't incurred this cost yet and are relatively cheap to recreate later if our load increases. Doing this is preferable to forcing ALL connections to load metadata at start, as this means the pool will always have a startup cost (in time), and the server will likely use more memory than really necessary.
Proposal: add new configuration parameter; if set, run the SQL string and get back a single-column, single-row result-set containing an arbitrary "value" as determined by the connection; prefer to close the lowest(?)-value connections first, to get down to the configured minIdle. Would require querying each open connection, when the pool is over-full, to determine which ones are most valuable. If not set, use current algorithm (based on age?).

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] [Updated] (DBCP-368) determine which connections to hold in pool by relative value

Posted by "Phil Steitz (Updated) (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/DBCP-368?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Phil Steitz updated DBCP-368:
-----------------------------

    Fix Version/s: 2.0
    
> determine which connections to hold in pool by relative value
> -------------------------------------------------------------
>
>                 Key: DBCP-368
>                 URL: https://issues.apache.org/jira/browse/DBCP-368
>             Project: Commons Dbcp
>          Issue Type: New Feature
>            Reporter: Philip Williams
>             Fix For: 2.0
>
>
> Situation: a Firebird-Classic database (each connection has independent metadata cache), in a metadata-heavy environment; many "quick" actions will never force the database server to load full metadata, but a few will, in our case resulting in 200MB of metadata cache in memory (takes about 3 seconds to load, visible latency to the user); any connection that has done this is "more valuable" to us, and we'd prefer to keep those connections in the pool, and remove idle connections which haven't incurred this cost yet and are relatively cheap to recreate later if our load increases. Doing this is preferable to forcing ALL connections to load metadata at start, as this means the pool will always have a startup cost (in time), and the server will likely use more memory than really necessary.
> Proposal: add new configuration parameter; if set, run the SQL string and get back a single-column, single-row result-set containing an arbitrary "value" as determined by the connection; prefer to close the lowest(?)-value connections first, to get down to the configured minIdle. Would require querying each open connection, when the pool is over-full, to determine which ones are most valuable. If not set, use current algorithm (based on age?).

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira