[postgis-users] MapServer PostgreSQL and persistent databaseconnections

Jan Hartmann jhart at frw.uva.nl
Fri Jul 11 02:58:17 PDT 2003


Paul, do you know WHY Oracle / SDE have such long connection times, 
compared to PostgreSQL? Is there some added functionality/bookkeeping 
invoved?

Jan

Paul Ramsey wrote:
>   The eternal question, to connection pool or not to connection pool. :)
> The reason why connection pooling is considered an important problem for 
> enterprise application developers is... because Oracle is really slow at 
> establishing a connection. Really, that is why! PostgreSQL connections times 
> are inconsequential. In the context of a process as computationally intensive 
> as Mapserver, PgSQL connections times might as well be considered to be zero.
>   This gets discussed every month or so on the Mapserver list, and the bottom 
> line is that 99% of your processing time for a given map creation will be 
> spent retrieving and rendering the data. Connection time for PostgreSQL is 
> completely ignorable in term of impact on system load. 
>   For Oracle and ArcSDE, connection time is non-trivial, and that is a 
> problem.
>   Because there are some connection time issues with some of the backends, for 
> Mapserver 4.0 there has been an enhancement to ensure that database 
> connections are pooled PER MAPPING INVOCATION. So within a particular 
> instantiation of the mapserv CGI, only one connection to each database is 
> used. So if I have 15 ArcSDE layers in my map, I will only connect ONCE, not 
> 15 times.
>   The infrastructure of inter-layer connection pooling exists in Mapserver 
> 4.0, and has been implemented on ArcSDE connections only, so far, but is 
> completely reusable for OracleSpatial and PgSQL layers. The implementation for 
> those layers just needs to be done.
> 
> Yours,
> Paul
> 
> Quoting Hubert Fröhlich <hubert.froehlich at bvv.bayern.de>:
> 
> 
>>Hi lists,
>>
>>I use UMN MapServer with PostGIS.
>>
>>When programming web applications with, say, PHP, Apache and PostgreSQL,
>>I can speed up database performance by keeping database connections
>>persistent and, therefore avoiding overhead by login and
>>authentification procedures.
>>
>>When using MapServer with PostGIS in a CGI-style way like in the
>>documetnation examples, I guess that I have to open and reopen the
>>PostGIS connection every time I start a mapserv process (at least I
>>haven't found a different way) because the database connection is
>>defined in the mapfile.
>>
>>Now my question: Is it possible that MapServer can control database
>>connections  and keep them persistent?
>>Or may be this an issue for the development team?
>>
>>Any hints will be greatly appreciated.
>>
>>Greetings,
>>
>>Hubert
>>
>>
>>--
>>
> 
> ------------------------------------------------------------------------------
> -
> 
>>Dr.-Ing. Hubert Fröhlich
>>Bezirksfinanzdirektion München
>>Alexandrastr. 3, D-80538 München, GERMANY
>>Tel. :+49 (0)89 / 2190 - 2980
>>Fax  :+49 (0)89 / 2190 - 2459
>>hubert.froehlich at bvv.bayern.de
>>
>>
>>_______________________________________________
>>postgis-users mailing list
>>postgis-users at postgis.refractions.net
>>http://postgis.refractions.net/mailman/listinfo/postgis-users
>>
> 
> 
> 
> 
> 
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users
> 




More information about the postgis-users mailing list