(RADIATOR) radiator performance benchmarks?

[Rodrigo Nuno Bragança da Cunha]

I'm sending attached a radius testing tool I've used in tha past to find 
various performance issues with our ouw radius servers.

It's a parallel test, and simulates a pool of clients making requests 
each X seconds and waiting at most Y seconds.

Requests are generated at a constant rate regardless of answers, and 
answers are processed as soon as they arrive. How badly serialization 
affects performance in case an authentication source has a delay (and 
other authentication sources have no delay) is imediatly apparent in 
this test. This is a tipical problem to solve with Fork() in Radiator, 
of course, in case you bump into something similar.

Should be easy to randomize the timers a bit and get a more "natural" test

You need a text file containing username/password pairs.

user1 password1
user2 password2
user3 password3
etc...

And the command line is someting like:
./radius-torture-test-v000.pl 500 user-list.txt result-file.txt 
radius-server:port 0.5 10

This means you are requesting 500 authentications from the file 
user-list.txt into radius-server:port and producing a log file called 
result-file.txt
The rate is 0.5 seconds (2 per second) and each costumer will wait for 
10 seconds at most.

A resonably fast machine is needed if you want many requests per second 
(dual-p3 1GHz works fine for 40-50 per second, perhaps much more).

Ops, I forgot one thing, the Secret. It's set direcly in the code at 
line 37.

Hope this helps!

--
Rodrigo

Tariq Rashid wrote:

>i'm going to start preparing next week to design a (possibly continuous)
>benchmarking and capacity planning system for radius rervers - inlcuding our
>radiator infrastructure.
>
>i realise that benchmarking is fraught with "dangers" - its easy to get
>numbers, hard to get meaning ful ones.
>
>can anyone point me to possible documentation or websites which advice on
>this topic. 
>
>regards
>
>tariq
>
>--
>Archive at http://www.open.com.au/archives/radiator/
>Announcements on radiator-announce at open.com.au
>To unsubscribe, email 'majordomo at open.com.au' with
>'unsubscribe radiator' in the body of the message.
>
>  
>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: radius-torture-test-v000.pl
URL: <http://www.open.com.au/pipermail/radiator/attachments/20040326/2f9d99d3/attachment.pl>