RedHat RHEL release to release comparison

Edit online

The purpose of this test was to determine how the WebSphere® environment performs on the various levels of RHEL. We wanted to compare RHEL 5.0 with RHEL 4.5.

For these runs we used Trade with the best tuning variations measured on our previous tests (see RHEL 4.5 Trade tuning variations).

Test case description

These tests used the Trade workload in our 3 tier environment with both systems, the WebSphere Application Server and the database, once on RHEL 4.5 and once on RHEL 5.0. We used the tuning variations measured on our previous tests (see RHEL 4.5 Trade tuning variations):

Increased network buffer count
Hardware checksumming

We ran tests using RHEL 4.5 and RHEL 5.0 to the DB2® for Linux® database. These runs set the baseline for future runs comparing different code levels.

DB2 database on Linux

Table 1. RHEL 4.5 AND RHEL 5.0 comparison
	WebSphere Application Server			Trade Results			DB2 Database
Distro	#CPU	%CPU	CPI	#Users	Resp	%ETR	CPU%	%CPU
*RHEL 4.5*	8	98.35	2.11	60	18 ms	100.0	4	40.57
*RHEL 5.0*	8	99.07	2.13	60	17 ms	100.3	4	47.5

This chart shows that both the RHEL 4.5 and RHEL 5.0 environments have throughput characteristics which are very close except for the DB2 for Linux CPU utilization. — Figure 1. RHEL 4.5 and RHEL 5.0 on a Linux database runs showing ETR, ITR, and CPU consumption comparisons

Observations

Both the RHEL 4.5 and RHEL 5.0 environments have throughput characteristics which are very close except for the DB2 for Linux CPU utilization. This means that when WebSphere Application Server and the database are on RHEL 4.5, it requires significant less CPU on the database for approximately the same throughput as RHEL 5.0.

Conclusion

From a performance perspective, the performance for the environment with WebSphere Application Server on RHEL 4.5 or RHEL 5.0 is the same. However, the database requires less capacity from eight CPUs when the environment runs on RHEL 4.5. At the moment it is not clear what causes the difference.