Random Access Memory (RAM), or memory, can cause performance problems if application memory requirements exceed the RAM capacity. This will cause the results to be swapped to disk.

Because the CONNX Client may be required to hold several record sets in resident memory to complete join or sorting operations, it has the largest memory requirement. To minimize memory usage when the result set is large enough, use disk based algorithms that sort in-memory pointers to the data.

The CONNX Listener uses very little memory and should never dominate resident resources.

CONNX Servers will batch record sets and may hold a number of resources they need to maintain the source database in a cursor state.