Even though multi-processors and memory are relatively inexpensive, CPU speed can still be a problem.

 

Raw computing power may be a problem depending on the application and expected result set size. Some CONNX operations (like sorting and aggregate operators) may be CPU intensive. Although CONNX core technology uses optimized algorithms, some operators must access each record no matter how large the data set.

 

Many organizations have multiple intensive applications running concurrently on a single machine. If you reduce the number of intensive applications concurrently running on a single machine (by adding more hardware to the network) processing should speed up.

 

On average, the CONNX remote servers use less CPU then the client. You can minimize some of the computing cost if you offload processing to the backend database. This is an internal CONNX design feature.

 

The CONNX Remote servers tend to be installed on legacy systems. There are many specific legacy CONNX hardware configurations such as RMS, DBMS and RDB running on VAX hardware running OpenVMS. Remember that the fastest VAX is much slower than today's average PC.

 

In some cases it may be possible to cluster machines or to use a Network File System implementation such that the data on a slow machine can be accessed from a much faster machine.

 

In a later section we will examine local versus remote ISAM access.