Nortel Symposium Agent, Справочное руководство

Страница: 33 / 38

June 2001 System configurations
Planning and Engineering Guide 33
CPU loading measurement data
Performance Monitor logs were collected during the testing for test cases 2
through 5. (The SA Server log file for Test Case 4 contained corrupted data and
therefore was not used. The corresponding TAPI log file was not corrupted and
could be used.) This data was analyzed and a simple model of Symposium Agent
CPU load as a function of call rate was derived.
Note: These calls were simple calls and had no other processing associated with
them.
The log data for the SA server and TAPI server are summarized below in the first
and second tables respectively:
Test Case Logfile Platform CPH CPU (%) CPU-Sec (in 1 hour)
2 NoLoad PII400 0 10% 0
3 LiteLd_02116 PII400 4000 12% 72
5 MLd_0216 PII400 8000 15% 180
SA Server CPU Vrs. Call Rate
y = 0.0216x
R
2
= 0.9842
0
50
100
150
200
0 5000 10000
CPH
CPU-Sec
CPU-Sec (in 1 hour)
Linear (CPU-Sec (in
1 hour))
Test Case Logfile Platform CPH CPU (%) CPU-Sec (in 1 hour)
2 NoLoad PII400 0 1% 0
3 LiteLd_02116 PII400 4000 2% 43.2
5 MLd_0216 PII400 8000 5% 144
4 HLd_0215 PII400 14000 8% 244.8