At 05:28 AM 5/23/01, Terence Lee wrote:
>Basically your network is already saturated. Any broadcast over 40% on
>ethernet is saturated.
40% of what? When stating percentages you need to also state what they are
a percentage of.
The original poster said the Linux machine is "generating 14% of broadcast"
which also doesn't make sense, but perhaps he is not a native English
speaker.
Is it 14% of the overall bandwidth utilization? Usually broadcast traffic
measurements are stated as a relative measurement. They are stated in
relationship to the amount of total bandwidth that is in use. If only 4% of
the total capacity of the network is in use, then 14% of that percentage
being used for broadcasts is not a problem.
A general rule of thumb is that no more than about 20% of your traffic
should be broadcasts, although the real answer is "it depends." The reason
to avoid broadcasts is because they interrupt the CPU on every device in
the broadcast domain. I wonder if this is still a problem with 1.7 GHz P4
processors!? ;-)
If you attach your fluke to a switched port, be aware that you will only
see broadcasts or frames address to that port, (unless you use SPAN).
Sometimes we get fooled into worrying about broadcast traffic because we
forget that we aren't seeing the non-broadcast traffic because of switch
behavior. Troubleshooting used to be so much easier in the days of shared
Ethernet. ;-)
First give some thought as to whether this "14% problem" is even worth
troubleshooting. If it is, you might want to get a protocol analyzer so you
can see what the Linux station is sending. Also check the configuration on
the Linux server. What services is it configured to be running? Could it be
running RIP and sending routing table broadcasts? What is this Linux
machine supposed to be doing besides acting as a Syslog server?
Regarding the "rule" that Ethernet utilization shouldn't exceed 40%,
remember that it was based on testing in a non-production network with
traffic patterns that had no resemblance to real-world networks. You need
to know what's connected to the Ethernet and what the connected stations
should be sending to know if the rule applies in your situation, which is
unlikely.
Check out this great paper by David Boggs, one of the creators of Ethernet
called "Measured Capacity of an Ethernet: Myths and Reality." The URL is:
http://www.research.compaq.com/wrl/publications/abstracts/88.4.html
Priscilla
>Your syslog box should not really be generation
>anything if you are just using it to log. Try using another nic to see if
>that helps
>
>--
>Terence Lee, MCSE, CCNA,
>
>""Sim, CT (Chee Tong)"" wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > Hi.. I used a fluke meter to check the health of our network. I found
one
> > of our Linux server are generating 14% of broadcast. But I don't know
how
> > to reduce it? any idea? For your information, we use this Linux box act
>as
> > a syslog server to receive message from all our switches. Does it
> > contribute to the broadcast? If the fluke meter found my network has 45%
> > broadcast, will it slow down my network?
> >
> > Chee Tong
> >
> >
> >
> > ==================================================================
> > De informatie opgenomen in dit bericht kan vertrouwelijk zijn en
> > is uitsluitend bestemd voor de geadresseerde. Indien u dit bericht
> > onterecht ontvangt wordt u verzocht de inhoud niet te gebruiken en
> > de afzender direct te informeren door het bericht te retourneren.
> > ==================================================================
> > The information contained in this message may be confidential
> > and is intended to be exclusively for the addressee. Should you
> > receive this message unintentionally, please do not use the contents
> > herein and notify the sender immediately by return e-mail.
> >
> >
> > ==================================================================
> > FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
> > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=5600&t=5517
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
