From: Marcus Jensen (marcus@pobox.com)
Date: Sat Aug 09 2003 - 09:58:05 GMT-3
Brian McGahan made an excellent post I was reviewing just now.
http://www.groupstudy.com/archives/ccielab/200306/msg01688.html
<snip>
1. The 'bandwidth percent' command does a relative bandwidth reservation
based on the current available bandwidth of that interface.
policy-map QOS
class FTP
bandwidth percent 50
!
interface FastEthernet0/0
service-policy output QOS
!
R1#sh queue fa0/0 | in Available
Available Bandwidth 75000 kilobits/sec
We are reserving 50% of the bandwidth on this interface for FTP. Since the
available bandwidth on the interface is 75Mbps, and we are doing a
percentage reservation, FTP is effectively guaranteed a minimum of 37.5Mbps
(50% of 75% of 100Mbps).
2. 'bandwidth [kbps]' is an absolute reservation, and is NOT based on the
available bandwidth of the interface.
policy-map QOS
class FTP
bandwidth 50000
!
interface FastEthernet0/0
service-policy output QOS
R1#sh queue fa0/0 | in Available
Available Bandwidth 25000 kilobits/sec
'bandwidth [kbps]' is an absolute reservation, and is NOT based on the
available bandwidth of the interface. You can see that after configuring
'bandwidth 50000' under the FTP class, only 25Mbps is now available on the
interface.
<end snip>
The part that confuses me is where bandwidth in kbps is not based on
available bandwidth. When using bandwidth percent, the available bandwidth
number does not decrease in the show queue output, but we know that is where
the bandwidth is coming from. With bandwidth kbps, it does decrease. But
when Brian said it is not based on available bandwidth, this caused me to
pause. I had always thought the 50000 is still coming from the available
bandwidth of 75000. But maybe it means 50000 is taken from the whole
interface? The results are the same in this case because of round numbers,
but which is it?
Thanks,
Marcus
This archive was generated by hypermail 2.1.4 : Tue Sep 02 2003 - 18:53:56 GMT-3