Compaq 108164-003 - ProLiant - 800 White Paper page 14

W P
HITE APER
(cont.)
. .
. . Table 12: PS User Scenario
. .
. .
Operation # Operation
. .
. .
1 Create a new UPD entry
. .
. .
. .
. .
. .
2 Get a personalized Web page
. .
. .
. .
. .
. .
. .
. .
. .
. .
3 Get a personalized Web page
. .
and update personalization
. .
. . information
. .
. .
. .
. .
. .
As mentioned, other user scenarios are possible and may lead to varied platform transaction
. .
capacities. An Excel spreadsheet has been provided in Appendix A to address this issue.
. .
. .
. .
PS Test Procedures
. .
. .
. .
Before starting any tests, UPDs were created to simulate user communities of 1,000, 10,000,
. .
100,000, and 1,000,000 people. Then, the following basic PS test procedures were used:
. .
. .
At SUT boot time, automatically start IIS. IIS logging is enabled.
. .
. .
. . Rename the UPD with the appropriate-sized user community to the active one used by PS.
. .
. .
Run perfmon on the SUT and have it log the processor, Active Server Pages, HTTP Service,
. .
memory, and User Property Database performance data to a file, as well as show a real-time
. .
. .
chart of the CPU utilization and the appropriate transaction rates. The Network Interface and
. .
Physical Disk may also need to be monitored, depending on configuration performance.
. .
. .
Start the tests running on the client systems. The WCAT collects performance statistics and
. .
. .
logs them to a file.
. .
. .
At the end of each test, collect the perfmon log from the SUT, the PS log file, and the client
. .
performance data file.
. .
. .
. .
Reboot the SUT and test for the next-sized user community.
. .
. .
All tests were done on isolated LANs that were inactive except for the test traffic. Tests were run
. .
with a five minute warm-up period and a 10 minute measurement period using WCAT.
. .
. .
. .
PS Test Results
. .
. .
. .
Figure 3 summarizes the ProLiant 2500 PS performance measurements shown in Table 13.
. .
Similarly, Figure 4 summarizes PS performance for the ProLiant 7000 based on the data in Table
. .
. .
14.
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
14
Operation
Frequency Comment
5% 5% of the users will be new, so
create a new UPD entry for
each of them.
91% Each Web page will consist of
an HTTP GET of the .asp file
and 4 HTTP GETs of .gif files.
Each .asp file accessed will be
8,290 bytes long and each .gif
file will be 5,480 bytes long.
4% This does the same HTTP
GETs as does Operation #2
and also updates 2 properties
in the UPD for the simulated
user.