CE/PE throughput and CPU load
The CE/PE and the DB2 database server were defined with 2 virtual guest CPUs each for this scenario.
The ECM workload was increased until the 2 CPUs on the CE/PE guest are fully used.
The figure above shows the normalized transaction throughput for different workload levels and the CPU loads on the CE/PE and DB server. A throughput rate of '1' corresponds to the transaction throughput for the lowest workload level with one CE/PE guest CPU.
The blue (dark) line in chart shows the CE/PE transaction throughput for the increasing workload. The red and yellow (light) lines show the corresponding CPU load on the CE/PE and the database (DB) guest.
Observations
The workload was increased until almost 2 CPUs on the CE/PE guest are fully used. The load on the database server reaches 0.5 IFLs at the highest workload level. The resulting transaction throughput nearly quadruples at the highest workload level compared to the lowest workload level.
The throughput scales linearly until the CE/PE CPUs are fully used, resulting in a similar trend line for the throughput and CE/PE CPU load. This is a very good scaling behaviour, only the number of CPUs on the CE/PE guest limits the transaction throughput for the users. There is no indicator for any overhead at the higher workload levels, because the CPU load follows exactly the throughput and does not increase faster.