On Wed, Mar 13, 2013 at 12:52 PM, André Warnier <a...@ice-sa.com> wrote:
> Rainer Frey wrote: > >> On 12.03.2013, at 17:14, Christopher Schultz < >> ch...@christopherschultz.net> wrote: >> >>> On 3/12/13 7:54 AM, amit shah wrote: >>> >>>> I am using Oracle. Oracle JDBC Driver provides the Oracle >>>> Universal Connection Pool (UCP) which includes this feature< >>>> http://docs.oracle.**com/cd/E11882_01/java.112/**e16548/fstconfo.htm<http://docs.oracle.com/cd/E11882_01/java.112/e16548/fstconfo.htm> >>>> >of >>>> >>>> >>>> connection failover but since we use tomcat jdbc connection pool we >>> >>>> cannot use UCP. >>>> >>> Why not? >>> >> >> Because it would be two-level pooling? >> >> Also UCP has lot of synchronized code which leads to blocking >>>> threads and less concurrency support. >>>> >>>> Let me know your suggestions/thoughts. >>>> >>> I'm thinking that a low-performance fail-over is preferable to a >>> zero-performance non-fail-over. >>> >> >> Well, low overall performance, but possibility of failover in the >> hopefully rare case, >> may not be acceptable compared to high(er) overall performance and a >> search for other ways >> to perform failover. >> >> > I am not sure that I totally follow the arguments here, but it seems that > there exist > - a JDBC pool with good performance, but no fail-over > - a JDBC pool with failover, but bad performance > Why not move the problem one level higher, and instead of using one tomcat > with several pools, use several tomcats each with their own pool ? > Several tomcats can be configured as a "failover pool" of tomcats, no ? > > I am using the tomcat jdbc pool independently by placing the jars in my web-application. We use glassfish as our application server. So what you suggest is not an option. > > > ------------------------------**------------------------------**--------- > To unsubscribe, e-mail: > users-unsubscribe@tomcat.**apache.org<users-unsubscr...@tomcat.apache.org> > For additional commands, e-mail: users-h...@tomcat.apache.org > >
