Compare DB2 Connect to non-DB2 Connect

In this test, the IRWW workload driver user sent all transaction requests directly to DB2® on z/OS®, bypassing DB2 Connect™. This test was run with 500 users and think times of 2000 and 2500 milliseconds. We also ran 100 users with a think time of 100 milliseconds. Two virtual CPUs were defined for the DB2 Connect z/VM® guest and 1 GB of memory was used. The z/VM LPAR was assigned six CPUs and the z/OS LPAR was assigned four CPUs. The parameter, MAXAGENTS, was set high enough that the number of active connections was not limited.

Figure 1 compares the DB2 Connect runs to the non-DB2 Connect runs.

Observations

At a think time of 100 milliseconds and the lower number of users, bypassing DB2 Connect resulted in 30% higher throughput. As the think time and number of users increase, the difference in throughput decreases.

Conclusion

In the IRWW workload case, it appears that DB2 Connect has a higher impact on throughput than in the Trade 6 workload case, especially with a lower number of users. Bypassing DB2 Connect resulted in 6% to 30% higher throughput. The impact of DB2 Connect is higher - the shorter the think time. This is expected because here the request rate is higher and DB2 Connect has more activity to manage.