CONNX Data Integration Suite 14.8.0 | Enterprise Planning Guide | CONNX Performance | Input and Output
 
Input and Output
Disk I/O
Disk access has traditionally always been the bottle neck for database related systems. This includes retrieval and updating of physical data, storing temporary result sets, and logging. For more information, see Estimating Throughput Requirements.
Network I/O
Usually local LAN access with Ethernet is not a bottleneck.
The system loopback device is an internal CONNX feature. If you use the system loopback device, it can speed up component communications. If you plan to use the loopback interface, install the components locally.
Depending on your application's scale, you may get the best results if you create a distributed setup where the database resides on a different machine than the CONNX data access components.
Due to system resource thrashing, overall performance may decrease if too many applications are competing for system resources.
For more information, see Estimating Throughput Requirements.
Connection Pooling
Enable connection pooling to speed up the establishing of a connection. Connection pooling is much faster because many of the internal objects needed to facilitate a connection have been allocated and initialized. The JDBC/RCI server has connection pooling enabled by default.
The CONNX Remote server is spawned for the life of a connection. Normally, when the connection ends, the remote server drops away. For pooled connections, the remote server does not drop away but none of the original SQL processing context remains.
For more information on connection pooling, refer to the CONNX User Guide.