Within the past week, the developers of a new JBOSS application have run some batch tests of their application on our z9 ifl (the Linux guest is configured with 6 GB memory, 2 cpus, SLES 10 sp1, Apache JBOSS 4.2.2, IBM JAVA 1.6) as well as on a IBM Blade server VM (the Linux machine is configured with 4 GB memory, 2 cpus, RHEL 5 64 bit, Apache JBOSS 4.2.2, SUN JAVA 1.6)
On the IBM blade server, their batch process test took about 3 hours, on the z9 ifl, it took 16 hours. So it causes us some concern. In the ifl performance data, I am seeing paging in the double digits. Also I am seeing the single java process on this guest start in the high 70's percentage, then it increases over time to over 160% in the middle of the night after 13 hours, then it drops down to the high 70's. The other guests on the ifl are very quiet up until the time period after 5 hours, when a another zLInux configured the same way (except only one cpu) started having its java process start up run with over 80% cpu. At that time I seeing the 3 z/M ifl cpus going from using 30-50 percent utilization each (with most of it in emul time) to all three cpu's start using mid to high 80% utilization each and stay that way until the test ends. The reason we gave the one zLinux guest 2 pus for this test was because we had seen the java process for this application use a lot of cpu . I would appreciate any words of wisdom on ways to cut the batch processing time on the zLinux guest to be comparable to the blade server. Thanks Ken Spracklen ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
