From: ron.wilkerson@gmail.com
Date: Thu Feb 19 2009 - 10:49:07 ARST
Your example is a bit extreme. If the requirement is to provide (source) destination based load sharing, you should not tell the proctor that the task is impossible.
If that was the task, they would make sure that you didn't enable per packet load sharing and that you had all the available routes in the routing table.
The lab does not ask you to do something that is impossible.
Ron
-----Original Message-----
From: "Antonie Henning - MWEB" <AHenning@mweb.com>
Date: Thu, 19 Feb 2009 13:27:15
To: Nadeem Ansari<nadeem.ansari574@gmail.com>; ALL From_NJ<all.from.nj@gmail.com>
Cc: Ruhann<groupstudy@ru.co.za>; <ccielab@groupstudy.com>
Subject: RE: Packet switching
Caught this thread a bit late. In cases where the cisco documentation
has been proven wrong in the real world, play along with what the cisco
documentation says during the lab. I had two of these cases on one lab.
E.g if the question asked to ensure cef load balances per destination
over two links, explaining to the proctor that it is not strictly
possible because CEF will create 16 hash buckets based on the source and
destination will be fruitless.
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Nadeem Ansari
Sent: 19 February 2009 12:17 PM
To: ALL From_NJ
Cc: Ruhann; ccielab@groupstudy.com
Subject: Re: Packet switching
Hi Rauhann,
I also done lots of exercise regarding load sharing by router and found
that
CEF by default do per source/destination load balancing and process
switching do per packet load balancing
Mr. Doyle and You both are correct.
Hi Experts,
But I also faced one strange scenario where you have four paths to a
destination network and two path of them are configured with "no ip
route-cache" to enable process switching and remaining two are CEF
enabled,
in this case what will happen CEF behavior will take over or process
switching behavior will take over, As per my observation CEF behavior
will
take care of load balancing, Need your aopinions
Regards
Nadeem
On 2/17/09, ALL From_NJ <all.from.nj@gmail.com> wrote:
>
> Hello, hope all is well with you.
>
> The link to the IP switching section is:
>
>
http://www.cisco.com/en/US/docs/ios/ipswitch/configuration/guide/12_4/is
w_12_4_book.html
>
> There are some interesting options for these commands, cef in
particular.
>
> IMO, you do not need to memorize any of this, the docs seem pretty
clear
> for
> this and as long as you know where to find this info, should be a few
easy
> points.
>
> This link will help, although it probably will not answer all your
> questions. It is pretty helpful though.
>
>
>
http://www.cisco.com/en/US/products/sw/iosswrel/ps1828/products_tech_not
e09186a00801e1e46.shtml
>
> HTH,
>
> Andrew Lee Lissitz
>
>
>
>
> On Tue, Feb 17, 2009 at 5:31 AM, Ruhann <groupstudy@ru.co.za> wrote:
>
> > Hi
> >
> > Guys, please correct on the following
> >
> > In the event of load-sharing, ie one router with 2 link to another
> > By default routers running IOS 12.4 and later have CEF enabled by
default
> > on
> > most interfaces.
> >
> > Default CEF will use per source/destination load-balancing,
> > CEF per packet can be enabled with "ip load-sharing per packet"
> >
> > Then, you have fast switching and process switching:
> > Fast Switching (enabled with no ip cef and ip route-cache) does per
> > destination load-balancing only
> > when Process Switching (enabled with no ip route-cache) does
per-packet
> > load-balancing again.
> >
> > Please verify if above is correct, reason I asking, is I have found
more
> > descrepencies with the DOC-CD,
> > claiming default cef does per dst only, but mr Doyle says it does
per
> > src/dst.
> >
> > (I have come to trust the Mr Doyle's teaching over the Doc-Cd, even
> though
> > his articles any not only cisco related.)
> >
> > PS I did not mentioned here is the switching mechanism on the input
> > interface, as I am only after the differences conceptually.
> >
> > Regards
> > --
> > <ruhann>
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> >
This archive was generated by hypermail 2.1.4 : Sun Mar 01 2009 - 09:44:12 ARST