Re: : Cisco- Queing Algorithm
Martin Heusinger <MHeusinger <at> gmx.net>
2004-07-02 05:48:59 GMT
Hi there!
First in a single CPU router there are two queueing mechanisms: hardware Q
(tx-ring, always fifo) and a software queue (fifo, wfq, pq, ..., cbwfq,
llq). The software Q is only invoked, when the hardware Q is full. In
other words: Why should one Q a packet, if one can send it with line speed!
With respect to this you have a "on"-"off" behaviour of the software Q.
This is especially important for test scenarios: make sure you overload
the link or you´re NOT testing the software Q!
On the other hand there are also platform dependant queueing mechanisms.
F.e. the GSR has queues before the crossbar switch; some catalyst switches
have input Qs and a different structure on output. Make sure you check
your manuals (or visit the QoS course, where mainly everything is
explained in detail ).
Hope this helps
Martin
On Thu, 01 Jul 2004 19:08:05 +0000, M. ELK <elkou141061 <at> hotmail.com> wrote:
>
> Found the following in CSCO doc :
>
> quote
>
> Queing Algorithm are activated only when a device is experiencing
> congestion,and in most cases are deactvated only when congestion clears.
>
> Unquote
>
> Q1- What is the definition of "congestion" . ie: when the box will
> consider the link is congested ??.
> Q2- Not clear this "on" "off" behaviour . my understanding that the
> queing is done in hardware and
> should be always "on" irespective of the link congestion .
>
> Appreciate to explain .
>
> Brgds
>
> _________________________________________________________________
> STOP MORE SPAM with the new MSN 8 and get 2 months FREE*
> http://join.msn.com/?page=features/junkmail
>
> -------
> The MPLS-OPS Mailing List
> Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml
> Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
>
-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml