Sorry, DOE -> Design of Experiment. -----Mensagem original----- De: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] nome de Rivalino Matias Jr. Enviada em: quarta-feira, 7 de fevereiro de 2007 22:33 Para: [email protected] Assunto: RES: RES: [Ntop] Dropped Packages
Gary, that looks good ! Of course the download test suggested isn't too much accurate. It was just to get a feeling. But, if he wants to do a very accurate stress test he should to do a design of project (DOE). In this case, all input factors are combined and controlled in order to have a sample of output values that can be analyzed statistically. Also, in order to define those input values (e.g. packets size, bandwidth and others) it is very important to characterize the target environment the tested box will be running. On the other hand, if he is interested in a pure stress test, just to find the maximal capacity of his system, it is not necessary to characterize the real environment, but just to define the main protocols the box will be handling. I think in both approaches iperf is a great tool to use. Regards, Rivalino -----Mensagem original----- De: Gary Gatten [mailto:[EMAIL PROTECTED] Enviada em: quarta-feira, 7 de fevereiro de 2007 20:54 Para: [email protected]; [EMAIL PROTECTED] Assunto: RE: RES: [Ntop] Dropped Packages Yup. When using Iperf with UDP you can set the bandwidth you use. You can alsa control datagram/packet/frame size. So, you could test at 2Mb/s with 1500 byte packets, or 20Mb/s using 64 byte packets. From my experience a 500 byte packet size is a decent "average". I usually test a range of sizes - 64byte, 500byte, and 1500bytes... Gary -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Rivalino Matias Jr. Sent: Wednesday, February 07, 2007 6:43 PM To: [email protected] Subject: RES: RES: [Ntop] Dropped Packages -----Mensagem original----- De: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] nome de Philipp Boksberger Enviada em: quarta-feira, 7 de fevereiro de 2007 18:36 Para: [email protected] Assunto: Re: RES: [Ntop] Dropped Packages > Try to look at throughput statistics of your WAN traffic. Also, you could do > a stress test of your debian box in order to identify the maximal throughput > you can obtain. With a 2Mbit uplink would it be ok to reach a maximal throughput of 2Mbit? ==> I think so. This is what I get when I download a large file ad midnight, when nobody else is using the cable. But this doesnt sound like astress test to me. How can I perform such a test? ==> In the above explained case you are limited by the WAN link capacity. In my last msg I suggested you to do a stress test in order to identify the maximal capacity of your FW box. To do that, you may execute the same download test but putting the box directly connected to the file server and the client using a fast/gigabit ethernet. Regards, Rivalino _______________________________________________ Ntop mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop =========================================================================== "This email is intended to be reviewed by only the intended recipient and may contain information that is privileged and/or confidential. If you are not the intended recipient, you are hereby notified that any review, use, dissemination, disclosure or copying of this email and its attachments, if any, is strictly prohibited. If you have received this email in error, please immediately notify the sender by return email and delete this email from your system." _______________________________________________ Ntop mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop _______________________________________________ Ntop mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop
