On Wed, Mar 5, 2014 at 3:15 PM, S Ahmed <sahmed1...@gmail.com> wrote:
> Hi, > > With jdbc pool, is each socket connection in the pool handled by a separate > thread? > > Ahmed, thanks for asking this question - it is sometimes very confusing with all different kind of pools: connection pools, threadpools, etc... Chris pointed out already - the connection pool does not have any threads... It is not a process that runs in the background, these are just connection objects that are sitting in memory. Threads are created by the Tomcat container (executor) once the connection is received by the Connector. The created thread is then going to be handled by the container and it will go through the stack call, through Valves, Filters, Servlets, your middleware layer, DAOs, JDBC/JPA calls and then finally through your datasource object, connection, PreparedStatement, ResultSet, etc... and back all the way to the socket that browser initiated, returning the thread to the threadpool (e.g. http-bio-8080). Now, I am not sure about the connection pool implementation details, how connection pool keeps connections open, if there are any background threads that are handling connection management (closing abandoned connections, opening new connections as the demand rises, etc...) Maybe someone can comment on that. > Say you have 20 connections set to be open at minimum, does that mean there > will be 20 threads? If not, then there is a degree of serialization then > right? > Well, there will be no serialization, but rather "synchronization" as there are at most XXXX available connections, if the connection pool (datasource) reaches maximum allowed connections, depending on the implementation - it would be a blocking call, until the pool has available connection to provide... Look at the documentation for Tomcat default connection pool implementation: https://tomcat.apache.org/tomcat-7.0-doc/jdbc-pool.html Hopefully that clears some of the confusion. Cheers! Neven
