performance report?

[Anders Holm]

Hmm..

"Our tool" .. What tool is this? Something you built yourselves? Commercial
offering perhaps?

I've done performance testing, load testing and all osrts of other types of
testing. You say you have figures showing something, but you seem unsure why
the figures tell you waht they do. Then either you've got the wrong figures,
or not understanding what they actually tell you. More than likely you have
the wrong figures. Have you gotten data points for *all* involved parts? Not
just CPU, but network, for all involved systems?

What you probably percieve as a "problem" on the server side is one of 3
things:

A/ Your clients simply cannot produce more traffic to make the server break
a sweat. My money lies here.
B/ You've saturated the network between clients and server. Unlikely without
causing other side effects.
C/ You're using a database type backend (SQL or LDAP) and it is not keeping
up.

Without knowing what your setup is, what tools you're using, how those tools
are configured, how your clients and RADIUS (FreeRADIUS of course) server is
configured and what the claimed figures are, what level of help would you
concieve we could bring you?

Facts speak louder than anything. So far those facts are elusive. Shows us
your data, while also expecting that if you're using a commercial tool
(potentially written by yourselves) that vendor should be involved. If the
volume of data you're willing to present is fairly large, I'd opt myself to
add it to a web site somewhere and just present a link to the mailing list.

CPU utilization is not the sole metric. Performance testing is usually aroun
the same level as statistics, lawyers etc. It'll only show you one part of
"reality", usually a fabricated one which only works within certain set
parameters. Change one parameter and all your performance testing is null
and void.

Also note, one protocol behaves one way, another protocol behaves another
way. "Success" is measured differently depending on goal at hand.

//anders

2008/8/20 Kevin J <kevinsmailer at yahoo.com>

> Well, Radius protocol is not just machine-to-machine issue.  I think you
> don't understand how request protocol can be simulated by hammering with our
> tool.  We have tested various protocols by this tool.
>
> Per our test results, radius can reach the limit of requests by hammering
> easily but CPU was still low. We have various statistics on all these.  My
> point is that radius was not able to use full cpu resource until reaching
> max number of handful requests.
>
> Your point with more clients does not make sense because we already reached
> max reqeusts hammering by our tool and that was same regardless of adding
> more clients under multi-threaded enviroment.
>
> ----- Original Message ----
> From: Anders Holm <anders.holm at sysadmin.ie>
> To: FreeRadius users mailing list <freeradius-users at lists.freeradius.org>
> Sent: Wednesday, August 20, 2008 12:52:20 PM
> Subject: Re: performance report?
>
> I still do ...
>
> I've had 10 multi core boxes hammering one server, still not enough .. You
> need more clients .. ;) RADIUS as such requires very little from the server
> side in terms of CPU. All it really does is compare x with y and then
> respond yes or no, once you strip down all the various variants of auth
> protocols. That's not a high requirement. I'm confident if you use a SSL
> enabled protocol, your CPU on the server is spending more time per request
> doing the necessary SSL stuff than RADIUS related work ..
>
> A pint of unspecified beverage says you'll need more client CPU .. I'll
> agree with the pint ..
>
> //anders
>
>
> On 20/08/2008 20:45, "Kevin J" <kevinsmailer at yahoo.com> wrote:
>
> Well, that's why I am saying we used the nas simulation tool.  We can
> hammer a lot of traffic with this multi-threaded tool and also we tried at
> least three client boxes so don't assume our traffic was not enough.
>
> ----- Original Message ----
> From: Anders Holm <anders.holm at sysadmin.ie>
> To: FreeRadius users mailing list <freeradius-users at lists.freeradius.org>
> Sent: Wednesday, August 20, 2008 12:25:19 PM
> Subject: Re: performance report?
>
> Re: performance report? It is not likely your actually putting too much
> strain on the server side. You'll need quite a lot of machines hammering the
> RADIUS server before it'll break into a sweat. The client side would have
> higher CPU utilization then the server side, per request.
>
> Comparing one program to another is not exactly comparing apples with
> apples. It's more like comparing a duck with a fork lift. One flies, the
> other just doesn't (or rather, when it does, you don't want to be there to
> see it) ...
>
> //anders
>
> On 20/08/2008 20:18, "Kevin J" <kevinsmailer at yahoo.com> wrote:
>
> Does anybody know the performance on Sun T-1000?
> Just noticed that radius cannot reach more than 20% CPU time when we ran a
> heavy traffic with nas simulations.  We have tested some other programs and
> could reach even more than 90% so just curious anybody experienced the
> similar result.
>
>
>
>
> ------------------------------
> -
> List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
>
>
>
> ------------------------------
> -
> List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
>
>
>
> -
> List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freeradius.org/pipermail/freeradius-users/attachments/20080821/75d32cd1/attachment.html>