Timesheet Benchmark 090518 on Translation Test Server
Reference:
Results:
- 1 parallel benchmarks running: 419.5 ms per page (0 seconds delay)
- 5 parallel benchmarks running: 1210.2 ms per page (0 seconds delay)
Results Interpretation:
- About 1 second load time per page is a very acceptable value.
- With a 0 seconds delay, each running benchmark corresponds to some 10 users concurrently trying to log hours.
- We can see that a normal office PC (4-core Pentium) is capable of serving 50 and more concurrent users trying to log their hours at the same time.
Server Hardware:
- Processor(s): 1 processor Intel Core2 Quad Q8200 @ 2.33GHz
- Ram: 4GByte
Server Software:
- Linux: CentOS 5.1 with kernel 2.6.18-128.1.6.el5
- AOLServer: 3.3oacs
- PostgreSQL: 8.2.11
Clint Configuration:
- Linux: CentOS 5.1
- Connected to the server via 100MBit Ethernet (same switch)
]project-open[ Configuration:
- ]po[ Version: V3.4.0.6.0
- Database: Translation Test System with a large real-world data set:
- 37456 Invoices
- 21824 Translation Tasks
- 12050 Projects
- 2093 Users
- 913 Companies
- debug = true
- ns_param maxconnections = 20
- ns_param maxthreads = 20
- ns_param minthreads = 5
- Disabled all but the Portlet components defined above
- Parameter: ShowLeftFunctionalMenupP = 0
Benchmark Configuration:
- Timesheet Benchmark 1.0
- num_repetitions = 10
- wait_between = 0 (no delay between loading pages)
Detailed Results per Page in ms (1 parallel benchmark):
- / - 75.7 (2x)
- /become - 42.9
- /intranet-timesheet2/hours/index - 133.7
- /intranet-timesheet2/hours/new - 2621.8
- /intranet-timesheet2/hours/new-2 - 115.0
- /intranet/ - 489.7
- /intranet/projects/ - 307.5
- /intranet/projects/index - 377.3
- /intranet/projects/view - 441.7
- /register/logout - 23.5